TRADING STRATEGIES PROJECT

Impact of expected & unexpected news in financial markets

Introduction
In this news: project, we have analyzed reaction of the markets to the following types of Reaction to positive surprises in expected news Reaction to negative surprises in expected news Reaction to positive unexpected news Reaction to negative unexpected news

As a primer, lets look at some of the very recent events of expected and unexpected news affecting the markets: 1. Reaction of the E-mini S&P500 futures to negative surprises in Ben Bernankes scheduled press release:

2. Strengthening of the Swiss Franc (CHF) and gold prices after the negative surprise of S&Ps downgrade of US credit rating: USD/CHF

XAU/USD

3. Reaction of BANK NIFTY index to negative surprise of RBI hiking interest rates to more than expected levels:

Literature Review
NBER: Stock Prices & Economic News (Douglas K. Pearce & V. Vance Roley) We find daily interpretations of various economic & financial news in the media. But there has been no quantitative model that provides systematic evidence on price movements. According to the Efficient Market Hypothesis, the current prices incorporate all existing and expected information. Thus, on arrival of news, any change in stock price can be attributed solely to unexpected news. It is also assumed that stock prices change instantaneously on arrival on any new information. Based on these hypotheses, Pearce and Roley built an economic model in an attempt to model price changes to unexpected information. In order to assess the impact of new economic news on stock prices, Pearce and Roley used the following model:

If we were to follow the EMH, each value in c should be zero as current stock prices reflect expected economic data and hence cannot contribute to change in stock prices. Since stock prices change immediately on arrival of new information, surprises that have occurred on previous day cannot contribute to change in stock prices on the current day. Therefore, each element of di is zero as well. So effectively, the model can be described as:

The key findings of this model were as follows: 1. Surprises related to monetary policies significantly affect stock prices 2. There was limited evidence of impact from surprise in inflation figures 3. There was almost no impact of surprises in economic activity 4. There was limited evidence of stock price responses to surprises beyond the announcement day In academic literature, we found that the economists were actually interested in the actual rise in price level to different news and finance professional were more interested in the volatility impact of the news. We summarize briefly below different academic articles that addresses these topics. Impact of Scheduled news on volatility Patell and Wolfson (1981), ex-ante and ex post effects of quarterly earnings announcement reflected in option and stock prices.

In this study, Patell and wolfson studied the effect of investors anticipation of impending informative disclosures on the behavior of option and stock prices. They observed the option prices prior to and immediately after the scheduled news announcements. They noticed that while the market expectancy suggests that the price of the stock should not be affected until the actual news is out and digested, the (implied) volatility of the stocks reflected in option prices is likely to increase as the investors are unsure and are anticipating the news. They noticed that the variability of the stock price would increase on the days of the announcement rising up to the actual announcement and decreasing after that. The empirical test verifies four interrelated phenomena Realized increase stock price variability at the announcement date Anticipation of increased stock variability evident in preannouncement option prices, i.e. ex-ante content Confirmation of anticipated information content reflected in announcement date option prices Ordinal correspondence between the magnitudes of the first two effects.

Nikkinen and Sahlstrm (2001), Global stock market reactions to scheduled U.S. macroeconomic news announcements. In this study, Nikkinen and Sahlstrm extend the previous study to observe the effects of U.S macroeconomic news announcements and its impact on various stock markets. The analyzed the GARCH volatilities around ten important scheduled macroeconomic news announcements on thirty five local stock markets in six regions. One interesting observation is the relation between a countries development status and the correlation with U.S. news. While the G7 countries and Asian countries were correlated to the U.S. economic data, transition countries like Latin America were not affected. The macroeconomic news announcements investigated are consumer confidence, consumer price index, employment cost index, employment situation, gross domestic product, import and export price indices, NAPM(National Association of Purchasing Management report): manufacturing and non-manufacturing, producer price index and retail sales. The local stock markets represent the G7 countries, the European countries other than G7 countries, developed and emerging Asian countries, the countries of Latin America and countries from Transition economies. In this paper, it is hypothesized that uncertainty associated with the announcements of the U.S. economic indicators is reflected differently in volatilities of local stock exchanges. This is to be expected, given that the stock markets differ considerably in terms of size, industrial diversity, and proportion of foreign ownership. The magnitude of reaction is therefore hypothesized to depend on the degree of integration and development of the particular market. The degree of economic integration affects stock market reactions in two main ways. First, it affects the performance of companies from small and medium-sized enterprises (SMEs) to large multinational companies (MNCs). For example, MNCs are not dependent on the situation on one particular market but the worldwide economic situation affects their performance. Similarly, the success of SMEs can either depend directly on the worldwide economic situation or indirectly, for example, through their multinational customers. Second,

stocks of local exchanges are owned by both local and foreign investors and the proportion of foreign ownership varies across different exchanges and over time. They also reinforced the findings of Patell and Wolfson (1981). They observe the jumps in implied volatility are sudden on the day of the announcement and do not last after that. This provides a significant finding that the effect of scheduled news is usually digested in a day. Impact of unscheduled news Fang, L. and Peress, J. (2009) Media coverage and the cross-section of stock returns In this study, Fang L and Peress J investigated the hypothesis that by reaching a broad population of investors, mass media can alleviate informational frictions and affect security pricing even if it does not supply genuine news. Mass Media Outlets, such as newspapers, play an important role in disseminating information to a broad audience, especially to individual investors. Every weekday, some 55 million newspaper copies are sold to individual readers in the United States, reaching about 20% of the nations population. If we consider online subscriptions and multiple readers per copy, the actual readership of the printed press is even larger, and certainly far broader than other sources of corporate information such as analyst reports. Given mass medias broad reach, one might expect it to affect securities markets. Interest in the relation between media and the market has been on the rise among both researchers and practitioners. They found that stocks not covered by the media earn significantly higher future returns than stocks that are heavily covered, even after accounting for widely accepted risk characteristics. A portfolio of stocks with no media coverage outperforms a portfolio of stocks with high media coverage by 3% per year following portfolio formation after adjusting for market, size, book-to-market, momentum, and the Pastor-Stambaugh (2003) liquidity factor. The return difference is particularly large among small stocks, stocks with low analyst coverage, stocks primarily owned by individuals, and stocks with high idiosyncratic volatility. In these subsamples, the no-media premium ranges from 8% to 12% per year after risk adjustments. Thus, the return premium for stocks with no media coverage is economically significant. They also provided empirical evidence of two phenomena because of which the above observation may be possible. First, it may be a liquidity-related phenomenon. Alternatively, the no-media premium may represent compensation for imperfect diversification. Tetlock, P. C. (2007) Giving content to investor sentiment: The role of media in the stock market, He quantitatively measures the interactions between the media and the stock market using daily content from a popular Wall Street Journal column. He finds that high media pessimism predicts downward pressure on market prices followed by a reversion to fundamentals, and unusually high or low pessimism predicts high market trading volume. Casual Observation suggests that the content of news about the stock market could be linked to investor psychology and sociology. However, it is unclear whether the financial news media induces, amplifies, or simply reflects investors interpretations of stock market performance. This paper attempts to characterize the relationship between the content of media reports and daily stock market activity,

focusing on the immediate influence of the Wall Street Journals (WSJs) Abreast of the Market column on U.S. stock market returns. First and foremost, He finds that high levels of media pessimism robustly predict downward pressure on market prices, followed by a reversion to fundamentals. Second, unusually high or low values of media pessimism forecast high market trading volume. Third, low market returns lead to high media pessimism. These findings suggest that measures of media content serve as a proxy for investor sentiment or non-informational trading. By contrast, statistical tests reject the hypothesis that media content contains new information about fundamental asset values and the hypothesis that media content is a sideshow with no relation to asset markets. This study systematically explores the interactions between media content and stock market activity. He constructs a straightforward measure of media content that appears to correspond to either negative investor sentiment or risk aversion. Pessimistic media content variables forecast patterns of market activity that are consistent with the DeLong et al. (1990a) and Campbell et al. (1993) models of noise and liquidity traders. High values of media pessimism induce downward pressure on market prices; unusually high or low values of pessimism lead to temporarily high market trading volume. Furthermore, the price impact of pessimism appears especially large and slow to reverse itself in small stocks. This is consistent with sentiment theories under the assumption that media content is linked to the behavior of individual investors, who own a disproportionate fraction of small stocks.

Regression of price change with earnings surprises

Listed companies in India and US report their earnings quarterly. Expected earnings of firms are already incorporated in the stock prices in accordance with efficient market hypothesis. Earnings estimates are obtained by taking average of the opinion from analysts belonging to different brokerage houses and are available on databases such as Bloomberg. However, prices of individual stocks show large fluctuations if the quarterly earnings reported by firms deviate significantly from the expectations in either negative or positive direction. Both institutional investors and retail investors take appropriate actions to factor in the new surprise information coming in. We have tried to analyze the fluctuations in stock prices near earnings results announcement dates. The data for earnings estimates and stock price movement was downloaded from Bloomberg for both Indian and US firms. However, the data for earnings estimates available for Indian firms was scanty since only yearly data starting from 2007 was available. Running regression on insufficient data would have led to inconclusive results. Therefore, leading US stocks were considered for analysis due to lack of sufficient data points for Indian stocks on Bloomberg.

Correlation between stock returns from T-5 to T and T to T+1

A correlation was obtained between stock price holding period returns for a period starting from five days prior to earnings announcement to the announcement date and

the returns on the date the earnings was announced. The following results were obtained. Stock Google IBM Agilent Yahoo HP Microsoft GE Correlation -0.192 -0.01 -0.051 0.097 -0.067 -0.242 -0.426

There is no significant positive correlation in the above data, which indicates that the stock returns after earnings announcements differed significantly from past returns. This proves the fact that insider information did not have a role to play and that earnings estimates were fairly accurate and the prices fluctuated due to genuine surprises in the reported earnings. Choice of independent variables (using Google) Estimates were available for a myriad of parameters such as Sales, EBITDA, Net Profit, and EPS. Returns on the day of results announcement were regressed separately with surprises in Sales, EBITDA, Net Profit, and EPS to find out the surprise variable which would best predict the stock price movement. Parameter Sales Surprise EBITDA Surprise Net Profit Surprise EPS Surprise All 4 Adjusted R square 0.0006 0.1122 -0.05811 0.1757 0.0755

Therefore, EPS was chosen as a proxy for earnings surprise since it has the highest predictive power. The same results were confirmed for IBM.

Regression results
Adjusted R square values for regression between percentage change in price and percentage surprise in EPS estimates was found to be low (0.176 for google and 0.123 for IBM). To our surprise, we found several data points where the earnings surprises

and price changes seemed to be moving in opposite direction. While the reported earnings significantly beat expectations yet the stock price moved down and vice versa. We tried to identify the extraneous factors which might be distorting our results. One possibility was that a company might be inflating its earnings artificially using corrupt accounting practices and thus analysts who read beyond the obvious might be shorting the stock causing it to take a beating in market. This was unlikely since most of the US stocks we considered were blue chip. Another possibility could be that the estimates available from Bloomberg might be flawed and that the analysts look at their own estimates obtained privately to take investment decisions. Chances of this occurring were also very low since Bloomberg can be regarded as a reliable source for US specific data. Therefore, it was highly likely that the broad market movement on a particular day might be the culprit. If the overall market is in a significant downtrend, even positive earnings surprises might not be enough to catapult a stock to higher prices. Regression approach Therefore, we set out to eliminate the effect of market movement (S&P 500 index) from our analysis. Two kinds of regressions were carried out. -Stock price movement was regressed with S&P movement and earnings surprises as the independent variables. -Excess return (Stock return S&P return) was regressed against earnings surprises. Regression results from IBM using both the approaches are as under:

Using market movement and EPS surprise As is evident from Figure 1, both EPS surprise and market movement have a p value less than 0.05, which indicates that both the variables are significant. The adjusted R square is 0.50.

Using excess returns Excess returns of a stock over the market are again significantly explained by EPS_Surprise at 95% confidence level, as is indicated by a small p value of 0.03. Results (adjusted R square) from regression A summary of adjusted R square values for different stocks under both kinds of regressions is as under. While results for google, IBM, Agilent and Yahoo are in line with expectations, the results from HP and Microsoft indicate that there might be other subjective factors also which might not be captured by our regression e.g. performance relative to peers. Using both S&P and Earnings Surprise 0.3027 0.5039 0.3137 0.093 0.1156 -0.013 -0.059

Stock Google IBM Agilent Yahoo HP Microsoft GE

Using Excess Return 0.1967 0.2027 0.0884 0.1086 -0.038 -0.001 -0.0614

Dividend announcements We also did regression analysis on dividends to find out how much of the stock return is influenced by the expected part of dividends and how much by the unexpected

(surprise) part of the dividend announcement. This analysis assumes the efficient market hypothesis which means that the present stock price is not affected by the past changes in the stock price. This would mean that the stock price at time t is dependent on information available only at that moment and not on any influence from data from t-1. Methodology The stock return on the day of a dividend announcement was regressed with the expected part and the unexpected part of the dividends to find out the influences of each on the stock return. 11 stocks from diverse sectors were taken (from Nifty) and their dividend history for the past 10 years was analysed. The stocks that were used for this are Airtel, Cipla, ICICI, ITC, Jindal Steel, JP Associates, L&T, NTPC, RIL, Tata Motors and TCS. The expected part of the dividend was assumed to be last years dividend multiplied with the GDP growth rate of that year. The unexpected part of the dividend was the actual dividend announce minus the expected part. The stock returns on the date of announcement were regressed with this data. Example Cipla Actual Dividend 100.000% 40.000% 100.000% 100.000% 100.000% 100.000% 100.000% 175.000% 150.000% 100.000% 107.222% GDP growth Expected rate Dividend 8.500% 109.100% 9.100% 0.000% 9.100% 104.900% 4.900% 109.800% 9.800% 109.300% 9.300% 109.300% 9.300% 189.525% 8.300% 162.600% 8.400% 103.800% 3.800% 110.925% Unexpected dividend -9.100% 40.000% -4.900% -9.800% -9.300% -9.300% -89.525% 12.400% 46.200% -3.703% Return 0.438% -3.246% -0.882% -0.275% 1.908% -2.333% 0.941% 2.480% -0.353% -0.147%

Final Interim Final Final Final Final Final Final Final Final Average

Regression results of Cipla

This was carried out for all the 11 stocks and their average was regressed with the average of expected and unexpected returns to get an overall view of how market reacts to expected and unexpected news. Results Stock Actual Dividend 20.000% 107.222% 100.000% 377.778% 179.2857% 17.250% 585.000% 16.500% 84.500% 108.500% 367.857% Expected Part 14.267% 110.925% 91.606% 305.551% 176.7511% 14.491% 578.368% 14.952% 75.348% 91.656% 341.145% Unexpected Part 5.733% -3.703% 8.394% 72.227% 2.5346% 2.759% 6.632% 1.548% 9.152% 16.845% 26.713% Return Correlation with expected -0.3321 0.0268 -0.0011 0.002 -0.0023 -0.2314 -0.0003 0.0305 0.013 -0.0008 -0.0016 Correlation with unexpected 0.3321 -0.0163 0.0078 0.0051 0.0064 0.1727 -0.0001 -0.0766 -0.0094 0.0172 -0.0023

Airtel Cipla ICICI ITC Jindal Steel JP Ass LT NTPC RIL Tata Motors TCS

-0.132% -0.147% -0.280% -0.296% -1.9965% -0.875% 2.716% 0.161% -0.942% 0.505% 1.384%

Conclusion Except JP associates, all stocks had correlation less than 0.1 indicating that the regression cannot be used to say much about the correlation of dividend announcements with stock returns. Most stocks (seven out of 11) are negatively correlated with the expected dividend part while 6 stocks are positively correlated with the unexpected dividend part Correlations with the expected and unexpected dividend parts had opposite signs for 10 out of 11 stocks indicating market views the expected and unexpected parts in opposite manner for any particular stock. The little influence that dividends have towards the returns are contributed by the unexpected part in most of the cases (8 out of 11)

Some Examples of news and its reaction in Indian market Case 1: Impact of a Macroeconomic news on Stock index The following is the sequence of news on RBI rate hike [26-Jul-11 07:47] RBI hints rate hikes [26-Jul-11 11:10] RBI sharply raises repo rate by 50 bps as inflation risk remains high [26-Jul-11 14:37] We were expecting a moderate hike of 25 bps but hiking rates by 50 bps is going to dampen the growth [26-Jul-11 17:12] IMC disapproves sharp hike in RBI's policy rates, says it will retard growth and abet inflation The Impact The news was expected but the extent of hike was more than expected which made the market fall on T, T+1, T+2, T+3 days. But on T+4 day the news was absorbed (other news were making effect) and NIFTY reversed.

Case2: Impact of Better than expected profit for a single company Headline : Mahindra Satyam jumps after turnaround Q1 result Date : 10-Aug-11 Mahindra Satyam jumped 9.85% to Rs 78.10 at 9:26 IST on BSE after the company reported consolidated net profit of Rs 225.18 crore for Q1 June 2011 compared with a net loss of Rs 327 crore for Q4 March 2011 . The results was announced after market hours on Tuesday, 9 August 2011.Meanwhile, the BSE Sensex was up 322.22 points, or 1.91%, to 17,180.13. The Impact NIFTY Close 9 Aug 2011 Price 10 Aug 1425Hrs 5072.85 2011 5139.95 NIFTY Ret 1.3227% Satyam Comp 71.05 78.7 Satyam Comp Ret

10.7671%

Clearly Satyam Comp market price had a positive impact of the news and the magnitude can be interpreted by the unexpectedness of the news. A high open indicates the over-reaction which was corrected instantly and again a 2nd phase of follower buying around 12:44Hrs was followed by a correction. Case3: Impact of announcement of new orders: Insider trading? [ 10-Aug-11 12:14] Suzlon announces 85 MW of new orders in India Suzlon Energy (SEL) has secured new orders worth Rs 483 crore for 85 MW (megawatt) in India from a broad range of customers in corporate, PSU and small/medium business segments. The key customers who have placed orders include GAIL India - Asia's leading gas utility company and a 'Navaratna' PSU of the Government of India. The 14.7 MW order from GAIL comprises seven units of Suzlon's S88 - 2.1 MW wind turbines to be commissioned in Gujarat within six months from the data of order. This is the second wind energy project order placed by GAIL with Suzlon. Other key orders include Khatau Narbheram & Co, Oswal Group, KRBL, and Varun Industries. The company made this announcement during the trading hours today, 10 August 2011. The Impact Buying in this particular scrip started in the morning and continued. A very small jump around the news publication Stock price fall after wards, although market continue to hold the price levels price trend reversal from a bottom of 41.8, below the Opening price and then started rising

Case4: Impact of possible dividend in a midcap stock

[10-Aug-11 10:51] Garnet International to consider final dividend Board meeting on 16 August 2011 The board meeting of Garnet International will be held on 16 August 2011 to consider and approve the audited results of the company for the year ended 31 March 2011 and to recommend final dividend, if any for the financial year ended 31 March 2011. The Impact Although dividend is uncertain the mere expectation boosted the price of this illiquid stock by 2.86% in 2 minutes The news came out at 10:51 but the price movement started at 10:53 (2 minute lag in diffusion)

Other Examples from Indian Market

According to efficient market hypothesis, price should adjust to the news information immediately and prices do not increase or decrease subsequently for the same news. But in actual market, reactions to news are not instantaneous. Market tends to overreact to bad news and under react to good news. High volatility in stock price after the news reaches market. Although the major part of price fall may happen during one session, it may take a couple of days for the market to completely factor in the news. Slow response to news can be explained by behavioral finance as conservatism exhibited by the investors.

Table showing price movement in some of Indian stocks due to + / - market news Price fall due to negative news Security Start Date Close Price End Date Close Price2 Trade Sessions % Price Fall News Satyam 06-Jan-09 178.95 09-Jan-09 23.75 2 87% Corporate Governace Issue GTL 16-Jun-11 408 20-Jun-11 127.95 2 69% Panic selling RCOM 10-Nov-10 179.45 25-Nov-10 136.9 10 24% 2G Scam SKS Micro Fin 12-Nov-10 919.85 18-Nov-10 639.4 3 30% Loan recovery issues Sun TV 31-May-11 390.05 02-Jun-11 272.55 2 30% 2G Scam Unitech 10-11-2010 89.8 19-11-2010 67.6 6 25% 2G Scam HCC 23-Nov-10 57.2 26-Nov-10 40.15 3 30% Hurdle for major project LIC Housing Fin 23-Nov-10 1308.75 26-Nov-10 931.15 3 29% Corporate Governace Issue MindTree 28-Jan-11 525.9 10-Feb-11 381.9 9 27% Founder quitting the board NIFTY 01-Aug-11 5,516.80 08-Aug-11 5,118.50 5 7% US Credit Rating downgrade Price rise due to positive news
Security Mahindra Satyam Hero Honda ICICI Bank Start Date 09-Aug-11 16-Dec-10 14-May-09 Close Price End Date Close Price2 78.35 1983 745 Trade Sessions 1 1 3 % Price Jump News

71.1 10-Aug-11 1696 20-Dec-10 536.55 19-May-09

Posted profit against 10% loss in last quarter 17% Unexpected jump in sales fall in interest rate to 39% boost lending growth

It can be observed from the above table that impact of the news is not instantaneous. For example in case of case of Satyam computers market took two trading sessions to react to the negative news of corporate fraud. In case of RCOM we find that the price decrease was gradual over a period of 10 trading sessions.