Do Swedish companies that exceed analyst earnings expectations experience abnormal stock returns beyond the initial effect on the announcement day? Ilaha Ashraf & Victor Larsson Abstract: This thesis aims to investigate whether Swedish companies that exceed analyst earnings expectations experience abnormal stock returns beyond the initial effect on the announcement day. The study focuses on firms listed on the Swedish OMXSPI index over two years. Employing event study methodology, daily stock and market returns were used to calculate abnormal returns (AR) and Cumulative Abnormal Returns (CAR) following quarterly earnings announcements. Earnings surprises were quantified as the percentage deviation between reported Earnings Per Share (EPS) and analysts' consensus estimates. Ordinary Least Squares (OLS) regressions were conducted to assess the explanatory power of earnings surprises over three-time horizons (5, 20, and 40 days). The findings reveal a statistically significant but weak relationship between earnings surprises and cumulative abnormal returns beyond the announcement day, particularly over 40 days (CAR-40). However, no significant relationships were observed for shorter time frames of CAR-5 and CAR-20. A portfolio strategy based on a long-short approach, designed to capitalize on earnings surprises, failed to deliver positive abnormal returns, instead underperforming relative to the market benchmark. These results align with the Efficient Market Hypothesis, suggesting that Post-Earnings Announcement Drift (PEAD) is less pronounced in the modern Swedish market. This study contributes to the existing literature by analyzing earnings surprises and PEAD in a Nordic context and highlights the limited profitability of exploiting earnings-related anomalies. It suggests further exploration of behavioral and structural factors influencing market inefficiencies. Bachelor’s thesis in Economics, 15 credits Fall 2024 Supervisor: Charles Nadeau Department of Economics School of Business, Economics and Law University of Gothenburg 1 Table of Contents 1 Introduction 3 1.1 Background 3 1.2 Problem Discussion 4 1.3 Aim and Purpose of the Study 4 1.4 Limitations 4 1.5 Outline 5 2 Literature Review and Theoretical Framework 6 2.1 Efficient Market Hypothesis 6 2.2 Behavioral Finance 7 2.3 Risk-Free Rate 7 2.4 Post-Earnings Announcement Drift (PEAD) 7 2.5 Momentum Investing 8 2.6 Market efficiency in different geographies 9 3 Methodology 10 3.1 Research Methodology 10 3.2 Data Collection 10 3.3 Sampling and Attrition 11 3.4 Variables 12 3.4.1 Abnormal Return (AR) 12 3.4.2 Cumulative Abnormal Return (CAR) 14 3.4.3 Earnings Surprise (ES) 14 3.4.4 Earnings Beat (EB) 15 3.4.5 Momentum 15 3.4.6 Profit 15 3.5 Statistical tests and models 15 3.5.1 Regression Analyses 15 3.5.2 Portfolio Strategy 16 4 Results 17 4.1 Descriptive Statistics 17 4.2 Regression Analysis 18 4.3 Portfolio strategy 21 5 Analysis and Discussion 22 6 Conclusion 26 References 27 2 1 Introduction Quarterly earnings reports are key events in the investing world, often triggering significant stock price movements, particularly when companies surpass analyst expectations (American Association of Individual Investors, n.d.). These immediate reactions raise a critical question: Does it signal lasting stock outperformance or merely a temporary, sentiment-driven response? According to the Efficient Market Hypothesis (Fama, 1970), stock prices should quickly reflect all available information, stabilizing shortly after the announcement. However, behavioral finance research, such as Bernard and Thomas (1989), challenges this view by highlighting post-earnings announcement drift (PEAD), where prices continue to adjust beyond the initial reaction. This study investigates whether Swedish firms that beat expectations experience sustained stock gains or short-term reactions. Given Sweden’s unique economic structure, this research explores market inefficiencies and investor behavior, offering insights into how markets incorporate new information. 1.1 Background The Swedish stock market provides a unique opportunity to study earnings surprises due to its blend of globalized industries and localized economic influences. Earnings announcements often trigger immediate stock price reactions, reflecting whether companies meet, miss, or exceed analyst expectations (Accounting for Everyone, n.d.). While earnings surprises have been extensively studied in larger markets like the U.S., the Swedish market remains unexplored. A recent study by Wiklund (2020) highlights the distinct characteristics of the Swedish stock market, including its smaller size, less liquid stocks, and concentrated industries. These characteristics may lead to unique stocks and concentrated industries, which may lead to unique post-earning price reaction patterns. The market's distinct industry focus, investor behavior, and regulatory frameworks may also create unique price reaction patterns. This study examines whether the Swedish market aligns with the EMH (Fama, 1970)- where stock prices adjust instantly to new information- or exhibits inefficiencies, such as PEAD (Bernard and Thomas, 1989). By analyzing delayed investor reactions, this research explores 3 whether earnings surprises lead to prolonged stock outperformance in Sweden, shedding light on local market dynamics and information processing inefficiencies. 1.2 Problem Discussion Although earnings announcements are critical to shaping investor expectations, the persistence of stock price changes following these announcements remains debated. In well-studied markets like the U.S., frameworks, such as the EMH (Fama, 1970) suggest price movements driven by new information should be instantaneous and short-lived. However, anomalies like PEAD, identified by Bernard and Thomas (1989), challenge this view, showing delayed price adjustments. Sweden’s stock market, with its smaller, less liquid stocks and concentrated industries, offers a unique setting where behavioral tendencies may amplify or diminish. This study explores whether earnings surprises lead to lasting stock price changes or short-term fluctuations, addressing a gap in understanding Swedish market behavior. The findings hold theoretical significance for market efficiency and practical value for Nordic investors. 1.3 Aim and Purpose of the Study The aim of this study is to determine whether Swedish companies that exceed analyst earnings expectations exhibit sustained stock performance beyond the initial post-announcement period. By focusing on the Swedish market, this research seeks to fill a gap in existing literature, where most studies on earnings surprises and stock performance are centered on the U.S. market. The findings from this study will enhance the understanding of market efficiency within the Swedish context, shedding light on investor psychology and the mechanism of price adjustments following earnings surprises. To guide the research, the following question is posed: Do Swedish companies that exceed analyst earnings expectations experience abnormal stock returns beyond the initial effect on the announcement day? 1.4 Limitations This study faces several limitations. It relies on publicly available financial data, excluding quantitative factors like management changes, geopolitical events, or market sentiment, 4 which may also influence stock performance. The focus on Sweden may limit generalizability to markets with differing regulatory structures, liquidity, or investor behavior. The analysis is restricted to a specific time period, which may not reflect long-term trends or evolving market conditions. Additionally, the reliance on analyst forecasts introduces potential bias, as forecast accuracy can vary, impacting the measurement of earnings surprises and subsequent stock performance. 1.5 Outline Chapter 1: Introduction- introduces the research question, background on earnings surprises, and the study’s relevance to the Swedish market. Chapter 2: Literature Review and Theoretical Framework- Reviews key theories like market efficiency, behavioral finance, and PEAD, with a focus on Swedish markets. Chapter 3: Methodology- Details the quantitative approach, data sources, and regression models used in explaining how the study controls for variables and ensures data accuracy. Chapter 4: Results- Presents the empirical findings, including statistical analyses of the impact of earnings surprise on stock performance over varying time horizons. Chapter 5: Analysis and Discussion- Interprets the findings in light of existing theories and compares them to prior research, particularly highlighting unique Swedish market behaviors. Chapter 6: Conclusion—This chapter summarizes the study’s findings, acknowledges limitations, and suggests areas for future research. 5 2 Literature Review and Theoretical Framework The literature review explores key theories and studies on the relationship between earnings surprises and stock performance. It covers the EMH, behavioral finance perspective on investor biases, and PEAD with a focus on market efficiency in Swedish markets. 2.1 Efficient Market Hypothesis The Efficient Market Hypothesis (EMH), proposed by Eugene Fama (1970), states that stock prices reflect all available information in three forms: ● Weak-form efficiency: Stock prices reflect all historical trading data. ● Semi-strong efficiency: Prices reflect all publicly available information, including financial statements and news. ● Strong-form efficiency: Prices include all public and private information. EMH is based on key assumptions such as investors rationality, information efficiency, and the absence of arbitrage opportunities. It suggests that, despite individual irrationality, collective investor behavior ensures efficient market outcomes. However, Fama (1991) acknowledges that anomalies such as the Post-Earnings Announcement Drift (PEAD) challenges the semi-strong form, indicating that stock prices may not adjust immediately to new information. This delay allows investors to exploit market inefficiencies and potentially achieve abnormal returns. Critics argue that EMH fails to fully explain recurring market anomalies such as speculative bubbles and investor overreaction (Shleifer, 2000). Behavioral finance researchers highlight cognitive biases such as overconfidence and loss aversion, which suggest that market participants are not always rational and challenge the core assumptions of EMH (Kahneman and Tversky, 1979). Understanding EMH’s applicability in the Swedish stock market is crucial for evaluating inefficiencies such as PEAD and assessing whether market participants react efficiently to new information within a Nordic context (Johansson and Persson, 2018). 6 2.2 Behavioral Finance Behavioral finance challenges EMH by incorporating psychological biases that influence investor decisions. Unlike EMH, which assumes rationality, biases such as loss aversion (Kahneman and Tversky, 1979) may contribute to market fluctuations that appear inefficient in the short run but could still align with long-term market efficiency. Additionally, biases like overconfidence and herding, contribute to phenomena like speculative bubbles, which contradict EMH’s assumptions of market rationality. Studies by Hede (2012) further emphasize how heuristics and cognitive biases contribute to financial anomalies, offering a robust critique of EMH. Building on these insights, Ding et al. (2024) investigate the impact of continuous good news on market volatility. Their study revealed that good news can unexpectedly increase volatility during bull markets as investors extrapolate expectations and overreact to positive trends. This behavior causes prices to deviate from rational valuations, highlighting the significant role of psychological biases in shaping market dynamics. Such findings provide further evidence of inefficiencies overlooked by EMH. 2.3 Risk-Free Rate The risk-free rate is typically represented by short-term government securities (e.g., 1-3 month Treasury bills), commonly used in empirical studies like the Capital Asset Pricing Model (CAPM) (Bodie, 1992). It serves as the baseline return for evaluating risky assets, which include investments such as stocks and corporate bonds that carry uncertainty in returns due to market volatility, firm-specific factors, and economic conditions. Fluctuations in the risk-free rate influence investment decisions and pricing models. Additionally, it plays a crucial role in pricing derivatives and bonds, as it impacts the discounting of future cash flow (Fama and French, (1996). 2.4 Post-Earnings Announcement Drift (PEAD) Post-earnings Announcement Drift (PEAD) describes the tendency for stock prices to continue adjusting in the direction of an earnings surprise for a period after the earnings announcement (Bond, 2023). This phenomenon challenges the EMH by suggesting that markets do not fully and immediately incorporate all available earnings information (Bernard & Thomas, 1989). 7 Ball and Brown (1968) first observed this drift noting that stock prices began adjusting to earnings information even before the official announcement due to accurate market expectations. Although the actual earnings release had a minimal immediate effect, prices continued to drift for about a month based on the earnings surprise. Building on this research, Bernard & Thomas (1989) further found that the drift could last up to 60 days, with the most significant price changes occurring within the first five days post-announcement. They also observed that small-cap stocks exhibited a more pronounced drift effect, especially when earnings surprises were larger, leading to abnormal returns of up to 4-6% over the 60 days. Earnings surprises, which drive Post-Earnings Announcement Drift (PEAD), occur when a company's reported earnings per share (EPS) differ from market expectations, typically based on analyst forecasts and market sentiment (Dechow et al., 2010). Positive surprises often lead to price increases, while negative surprises result in declines. However, several factors contribute to the delayed market response observed in PEAD (Kahneman & Tversky, 1979; Barberis & Thaler, 2003). The relevance of PEAD to this study lies in its implications for market efficiency, particularly in the Swedish stock market. If stock prices continue to adjust following earnings surprises, it suggests that the market may not be fully efficient in incorporating new information promptly. Studies such as by Hedberg and Lindmark (2012), have examined the presence and magnitude of PEAD within the Swedish market and provide valuable insights into whether investors can capitalize on delayed price adjustments and whether market anomalies persist in a relatively smaller and less liquid market compared to global counterparts. In conclusion, PEAD provides valuable insights into market efficiency by suggesting that stock prices do not always adjust instantaneously to new information. Analyzing this anomaly within the Swedish market context helps to determine whether persistent inefficiencies can offer potential opportunities for investors. 2.5 Momentum Investing Momentum investing is a strategy of buying high-return assets and selling low-return ones, often recognized as one of the most effective investment approaches (Barberis & Shleifer, 2003). This strategy is particularly profitable in small-cap stocks and those with low analyst coverage, with a pronounced effect on past losers (Hong, Lim & Stein, 2000). Momentum 8 investing, which involves capitalizing on past stock performance trends, is closely linked to PEAD. Studies by Chordia & Shivakumar (2006) explored the relationship between earnings surprises and momentum, finding that while the two are interconnected, they independently influence future returns. Their findings suggest that price momentum reflects earnings momentum, underscoring the pivotal role of earnings surprises in shaping stock price behavior. 2.6 Market efficiency in different geographies Chordia and Shivakumar (2006) also demonstrated that the return of portfolios, capturing PEAD, is significantly correlated with macroeconomic indicators like GDP growth, industrial production, consumption, inflation, and treasury bill returns. This suggests that the PEAD may vary across different countries and time periods. The efficiency of global stock markets and their impact on PEAD have also been explored. Giglio et al. (2008) assessed the relative efficiency of 36 stock markets worldwide, assigning scores from 0 (completely inefficient) and 100 (completely efficient). Efficient markets exhibit random price changes driven solely by new, unpredictable information. Their findings ranked the S&P 500 (USA) as the most efficient market with a score of 99.1%, followed by European indices like the DAX30 (Germany, 98.4%) and CAC 40 (France, 88.4%). Scandinavian markets, such as Denmark’s index (58.7%) were less efficient while emerging markets like China’s index scored 49.5% (Giglio, Matsushita & da Silva, 2008). 9 3 Methodology This section outlines the research methodology used in the study including a detailed explanation of the data collection methods, the variables utilized in the analysis, and the statistical tests selected to address the research question. 3.1 Research Methodology The study uses a quantitative research methodology, collecting data from Swedish publicly listed companies via S&P Capital IQ and Eikon Refinitiv. This method aligns with prior studies like Bernard & Thomas(1989), and enables relevant comparisons. The data spans from January 2022 to January 2024- a shorter time frame due to manual handling and time constraints. Statistical analyses, including descriptive statistics and regression analysis, were conducted using Stata. Cumulative abnormal return (CAR) is the dependent variable, and earnings surprise (ES and EB) is the independent variable, with the regression assessing stock performance following earnings surprises. To ensure data validity and reliability, samples from ten companies were cross-verified with financial reports. The study follows a deductive approach, testing existing theories against the collected data to evaluate their alignment with empirical results. 3.2 Data Collection The dataset is sourced from S&P Capital and Eikon Refinitiv, where data was retrieved from companies listed on the Swedish stock exchange. The data points for these companies were collected during the period from January 2022 to January 2024. More specifically, the data collected was the company's daily closing stock price from January 3rd, 2022, to 3rd January 2024. In addition to the company's stock prices, data was also collected on the Swedish stock market index OMXSPI closing price during the same time period. All this data was collected from Eikon Refinitiv. Data was also collected from the eight quarters that the companies reported during this period, covering the fourth quarter of 2021 to the third quarter of 2023. From these eight quarters, the reported earnings per share (EPS) were collected. From the same eight quarters, the analysts' consensus estimates for earnings per share (EPS) were also collected. The reason 10 why earnings per share was chosen, is because it has been used in previous studies (Bernard & Thomas, 1989; Livnat & Mendenhall, 2006). This makes our results more comparable. This data was gathered from S&P Capital IQ. All these quantitative variables were collected in the currency Swedish kronor (SEK). In addition to this quantitative data, information on the dates when the companies released their quarterly reports was collected from S&P Capital IQ. 3.3 Sampling and Attrition A selection process was carried out to obtain a robust dataset capable of effectively answering the research question. Since the study aims to investigate whether the post-earnings announcement drift exists on the Swedish stock exchanges, it was limited to companies listed in Sweden. The initial sample consisted of companies listed on Swedish stock exchanges. A total of 784 companies were listed at the selected time period. Subsequently, a refinement was made to only include companies that had analyst estimates for all eight quarters from the fourth quarter of 2021 until the third quarter of 2023. This was necessary because analysts' estimates are a key part when evaluating post-earnings announcement drift. Without them, the phenomenon cannot be evaluated. This refinement resulted in the removal of 636 companies, leaving 148 companies. The third refinement involved excluding companies that were missing stock price data in the selected time period. This was required because it is not possible to make the necessary calculations for a given company when the stock price data is missing. After this step, 13 companies were removed. The final refinement involved excluding companies that had other missing data points, which excluded 19 companies. This gave us the final dataset consisting of 116 companies, as seen in Table 1. 11 Table 1 Sample Companies Attrition Listed on a Swedish stock exchange 784 Has analysts estimates 148 636 Does not lack stock price data 135 13 Other missing data 116 19 Total attrition 649 This results in a final sample comprising 116 companies. With data spanning a two-year period, the dataset includes a total of 928 quarterly earnings reports and 58,696 daily stock price observations. 3.4 Variables Once the final dataset has been compiled, it is used to derive several key variables. These variables are central to the statistical analysis, as they represent the key factors being examined in the study. The selection and construction of variables are critical to ensure that the analysis accurately addresses the research question. In this section, we outline the various variables included in the analysis, their definitions, and how they are calculated from the raw data. These variables will then be utilized in the statistical tests to assess the relationships and patterns under investigation. 3.4.1 Abnormal Return (AR) To analyze the abnormal return for each stock we first calculated the daily stock return for each company and then the daily return for the index OMXSPI, over the two-year period. The stock return represents the percentage change in the stock price for each company from one trading day to the next. It was calculated using the formula: 𝑆𝑡𝑜𝑐𝑘𝑃𝑟𝑖𝑐𝑒 − 𝑆𝑡𝑜𝑐𝑘𝑃𝑟𝑖𝑐𝑒 𝑆𝑡𝑜𝑐𝑘𝑅𝑒𝑡𝑢𝑛 = 𝑡 𝑡−1 𝑡 𝑆𝑡𝑜𝑐𝑘𝑃𝑟𝑖𝑐𝑒 𝑡−1 This calculation was applied separately for each firm, ensuring that the stock returns were measured relative to the firm's own trading history. 12 A similar calculation was made with the OMXSPI, where the daily percentage change in the value of the OMXSPI index was measured. This served as a proxy for the overall market performance. The formula used was: 𝑂𝑀𝑋𝑆𝑃𝐼 − 𝑂𝑀𝑋𝑆𝑃𝐼 𝑂𝑀𝑋𝑆𝑃𝐼_𝑅𝑒𝑡𝑢𝑟𝑛 = 𝑡 𝑡−1 𝑡 𝑂𝑀𝑋𝑆𝑃𝐼 𝑡−1 Like the firm stock return calculations, the market returns were computed only when data for the previous day was available. While the OMXSPI index is the same across all firms, the calculation was structured within the same sorting framework to ensure consistency with the stock return data. By calculating these variables, we established the foundation for further analyses, such as the computation of the abnormal return. To evaluate the abnormal return of firms in relation to market movements, a series of calculations were performed. First, an ordinary least square regression was used for each stock. Their stock-specific coefficients alpha and beta were estimated to model the relationship between the stock returns and market returns. The regression equation used was: 𝑆𝑡𝑜𝑐𝑘𝑅𝑒𝑡𝑢𝑟𝑛 = α + β 𝑂𝑀𝑋𝑆𝑃𝐼_𝑅𝑒𝑡𝑢𝑟𝑛 + ϵ 𝑡 1 𝑡 𝑡 In this equation, alpha is the intercept capturing the average stock-specific return unrelated to market performance, beta is the sensitivity of the stock return to market movements. These coefficients were then used to calculate the expected returns for each stock on each trading day. The expected return for a stock on a given day was computed using the equation: 𝐸[𝑅 ] = α + β 𝑂𝑀𝑋𝑆𝑃𝐼_𝑅𝑒𝑡𝑟𝑢𝑛 𝑡 1 𝑡 The expected return represents the return expected based on the firm's historical relationship with market movements, as defined by the regression coefficients. With the expected returns calculated, abnormal returns were derived as the difference between the actual firm returns and the expected returns: 𝐴𝑅 = 𝑆𝑡𝑜𝑐𝑘𝑅𝑒𝑡𝑢𝑟𝑛 − 𝐸 𝑡 𝑡 [𝑅 𝑡] Abnormal returns are a measure of the firm's performance relative to what would be predicted based on market conditions. The abnormal return variables have been created in a similar way in previous studies (Bernard & Thomas, 1989; Livnat & Mendenhall, 2006). This 13 will make the result more comparable. Positive abnormal returns indicate that the firms outperformed expectations, while negative abnormal returns indicate underperformance. 3.4.2 Cumulative Abnormal Return (CAR) To evaluate the stock market's reaction to earnings announcements over different time periods, we calculated cumulative abnormal returns for three different event windows. The three event windows that were calculated were 5 days, 20 days, and 40 days following the earnings report. This was achieved by summing the abnormal returns, that were previously calculated, for each firm over the specified time periods. For each firm and each earnings event, abnormal returns were aggregated starting from the first trading day after the announcement, to avoid including the announcement day itself. The cumulative abnormal return for each time period was calculated as follows: 𝐶𝐴𝑅 = ∑ 𝐴𝑅 𝑡 The cumulative abnormal returns were calculated separately for each firm and earnings announcement date to ensure that the results were specific to individual events. In summary, the three variables that were created were: ● 𝐶𝐴𝑅_5 ● 𝐶𝐴𝑅_20 ● 𝐶𝐴𝑅_40 The three different time-specific variables were created to evaluate the stock market's reaction to earnings announcements over the short-, medium- and long-term. The cumulative abnormal return variable is measured over a shorter time frame compared to previous studies (Bernard & Thomas, 1989). This is because we aim to prevent the cumulative abnormal return variable from being influenced by the subsequent quarter. 3.4.3 Earnings Surprise (ES) To assess the extent to which a company’s actual earnings per share (EPS) deviated from market expectations, the earnings surprise was calculated for each earnings announcement. The earnings surprise measures the percentage difference between the firms' reported EPS and the analysts’ consensus estimate of EPS. This was calculated using the formula: 14 𝐸𝑆 = 𝐴𝑐𝑡𝑢𝑎𝑙 𝐸𝑃𝑆 − 𝐸𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑 𝐸𝑃𝑆𝐸𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑 𝐸𝑃𝑆 By dividing the difference between actual and estimated EPS, the calculation standardizes the magnitude of surprise, making it possible to compare across firms and events. This variable is also calculated similarly to previous studies to facilitate comparison (Bernard & Thomas, 1989; Livnat & Mendenhall, 2006). 3.4.4 Earnings Beat (EB) The earnings beat (EB) variable is a dummy variable that takes the value 1 if the actual EPS exceeds the estimated EPS and 0 otherwise. It is used to analyze whether the direction of the surprise matters more than the size of the surprise. 3.4.5 Momentum Momentum is a variable that represents the cumulative abnormal return (CAR) over the 20 days preceding the quarterly earnings report. This is a control variable that is included to help isolate the effect of earnings surprise since previous studies have found that both price momentum and earnings surprises influence future return (Chordia & Shivakumar, 2006) 3.4.6 Profit The variable Profit is a dummy variable that takes the value 1 if actual EPS is above 0 and takes the value 0 if it is not. This is a control variable that is included to evaluate and isolate the effect of a positive EPS on cumulative abnormal return after the report. 3.5 Statistical tests and models Here, the statistical tests and models underlying the study’s results are presented. The tests and models that will be used include regression analyses, as well as the presentation of descriptive statistics. Finally, a portfolio will be constructed to evaluate the performance of a trading strategy based on earnings surprises. 3.5.1 Regression Analyses A series of regressions were conducted to examine the relationship between cumulative abnormal returns and earnings surprises. The aim was to determine whether earnings surprises could explain variations in abnormal stock returns over different time horizons. 15 Specifically, the cumulative abnormal returns over three distinct periods were regressed on earnings surprises. The different control variables were also added to some of the regressions to see their effect. The regression models used was: ● 𝐶𝐴𝑅_5 = α + β 𝐸𝑆 + ϵ 1 ● 𝐶𝐴𝑅_20 = α + β 𝐸𝑆 + ϵ 1 ● 𝐶𝐴𝑅_40 = α + β 𝐸𝑆 + ϵ 1 ● 𝐶𝐴𝑅_40 = α + β 𝐸𝐵 + ϵ 1 ● 𝐶𝐴𝑅_40 = α + β 𝐸𝑆 + β 𝑀𝑜𝑚𝑒𝑛𝑡𝑢𝑚 + β 𝐸𝑆𝑥𝑀𝑜𝑚𝑒𝑛𝑡𝑢𝑚 + ϵ 1 2 3 ● 𝐶𝐴𝑅_40 = α + β 𝐸𝑆 + β 𝑃𝑟𝑜𝑓𝑖𝑡 + β 𝐸𝑆𝑥𝑃𝑟𝑜𝑓𝑖𝑡 + ϵ 1 2 3 3.5.2 Portfolio Strategy The portfolio strategy that is evaluated is a long-short strategy. The strategy evaluates the profitability of going long on firms with positive earnings surprises and short with negative surprises, excluding the announcement day to prevent potential biases. By calculating and aggregating abnormal returns, the strategy’s daily and cumulative performance was assessed. 16 4 Results This section presents the results of the empirical study. First, descriptive statistics are presented, followed by the regression models, and finally, the results of the portfolio strategy are presented. 4.1 Descriptive Statistics The table below presents descriptive statistics for the variables used in the various statistical tests. It provides information about each variable’s mean, standard deviation, maximum value, minimum value, and number of observations. The reason for presenting descriptive statistics first in the results section is to provide an overview of the data. Table 2 Variable Observations Mean Standard Deviation Min Max CAR_5 925 -0.0001 0.06 -0.291 0.325 CAR_20 925 0.001 0.101 -0.71 0.593 CAR_40 900 -0.003 0.133 -0.722 0.573 ES 924 -0.132 2.58 -56.14 22.65 EB 928 0.448 0.498 0 1 Momentum 900 0.062 0.184 -0.443 1.095 Profit 928 0.857 0.351 0 1 As can be seen in the table, all CAR variables contain 925 observations, except for CAR_40. The reason is that there was insufficient stock price data for certain quarters, and the same applies to the Momentum variable. The other variables contain 928 observations except for ES which contains 924, as four quarters had an estimated EPS of 0, making the calculation impossible. The mean values for CAR_5 and CAR_40 are both negative, while CAR_20 has a positive value, but all three are close to 0. This indicates that, on average, the stocks perform approximately as well as the benchmark the days after the report. 17 By looking at the other variables it could be seen that ES has a negative mean which suggests that, on average, actual earnings per share tend to be approximately 13% lower than estimated EPS. EB has a mean of 0.448, meaning approximately 45% have actual EPS exceeding the estimated EPS. The positive mean for Momentum measured as the cumulative abnormal return (CAR) 20 days before the quarterly report, indicates a slight upward trend in stock prices leading up to the earnings announcement. For the profit variable a mean of 0.875 indicates that in 87.5% of cases, the actual EPS is positive. 4.2 Regression Analysis Several regressions were performed to address the study’s research question. A total of six regressions were conducted, each including different variables. All regressions were analyzed at the 1%, 5%, and 10% significance levels. The first three regressions conducted had earnings surprise (ES) as the independent variable, and the different CAR variables as the dependent variable. This was done to investigate which time period had the greatest effect. Table 3 Dependent variable Adj R-squared = 0.001 CAR_5 Independent Variable Coefficient Std. Err. P-value Intercept -0.0001 0.001 0.905 ES 0.002 0.001 0.899 *** p<0.01 ** p<0.05 *p<0.1 Observations: 921 Table 4 Dependent variable Adj R-squared = 0.001 CAR_20 Independent Variable Coefficient Std. Err. P-value Intercept 0.001 0.003 0.797 ES -0.001 0.001 0.466 *** p<0.01 ** p<0.05 *p<0.1 Observations: 921 18 Table 5 Dependent variable Adj R-squared = 0.003 CAR_40 Independent Variable Coefficient Std. Err. P-value Intercept -0.003 0.004 0.496 ES 0.003 0.002 0.082* *** p<0.01 ** p<0.05 *p<0.1 Observations: 896 In the tables above, the results from the different regressions are presented. In the first regression, where the short and medium-term periods are analyzed, no significant relationship is found, and the explanatory power is also very low. In the third regression, a significant relationship between ES and CAR_40 can be observed at a 10% significance level and it is positive. This means that companies with higher earnings surprises also have higher cumulative abnormal returns over a 40-day period after the report. However, the relationship is very weak, and the explanatory power of the model is very low. Since CAR_40 was the only dependent variable that had a significant relationship, further analysis was conducted on that variable. In the regression shown in table 6 the variable EB is the independent variable instead of ES. This regression was conducted to evaluate if the direction of the surprise matters more than the size of the surprise. Table 6 Dependent variable Adj R-squared = 0.001 CAR_40 Independent Variable Coefficient Std. Err. P-value Intercept 0.003 0.005 0.659 EB -0.013 0.009 0.161 *** p<0.01 ** p<0.05 *p<0.1 Observations: 900 In this regression, it can also be seen that there is no significant relationship, and the explanatory power is very low. Since there was only one significant relationship between the variables CAR_40 and ES, further analysis was conducted specifically on these by adding control variables. 19 To examine the relationship between CAR_40 and the independent variables ES, Momentum, and their interaction term, a regression was conducted. The result can be seen in table 7. Table 7 Dependent variable Adj R-squared = 0.003 CAR_40 Independent Variable Coefficient Std. Err. P-value Intercept -0.002 0.005 0.746 ES 0.004 0.002 0.033** Momentum -0.027 0.024 0.257 ES x Momentum -0.021 0.016 0.201 *** p<0.01 ** p<0.05 *p<0.1 Observations: 896 Looking at the individual predictors, the variable ES had a statistically significant positive effect on a 5% significance level. This implies that a one-unit increase in ES is associated with a 0.005 increase in CAR_40 on average, holding all other variables constant. However, the effect is small, indicating a weak relationship. The other variables in the model do not have a significant relationship with CAR_40. Overall the relationships in the model are weak and the explanatory power is low. In the next regression, which result can be seen in table 8, CAR_40 was regressed on ES, Profit, and their interaction term. None of the coefficients in the regression had a significant relationship, which for example means that there is no relationship between firms that make profit and CAR_40. 20 Table 8 Dependent variable Adj R-squared = 0.001 CAR_40 Independent Variable Coefficient Std. Err. P-value Intercept 0.002 0.013 0.847 ES 0.004 0.002 0.101 Profit -0.006 0.014 0.654 ES x Profit -0.012 0.004 0.772 *** p<0.01 ** p<0.05 *p<0.1 Observations: 896 4.3 Portfolio strategy The chart below illustrates the cumulative abnormal returns for the long-short portfolio strategy over the period from January 2022 to January 2024. The strategy was designed to go long on firms with positive earnings surprises and short on firms with negative surprises, excluding the announcement day to mitigate biases. Chart 1 The graph reveals a consistent decline in abnormal returns over time. Starting in January 2022, the portfolio's abnormal return decreased steadily, reaching -8.3% by the end of the observed period. This downward trend suggests that the portfolio underperformed relative to the market benchmark. The results indicate that the proposed long-short strategy failed to generate abnormal returns over the observed period. 21 5 Analysis and Discussion This study's findings provide a comprehensive analysis of the Post-Earnings Announcement Drift (PEAD) phenomenon within the Swedish stock market. It evaluates its implications through the lens of market efficiency, behavioral finance, the risk-free rate, and trading strategies and compares market efficiency in different geographical regions. The study sought to answer the question of whether Swedish companies that exceed analyst earnings expectations experience abnormal stock returns beyond the initial effect on the announcement day. The regression analysis conducted in this study revealed a weak but positive relationship between earnings surprises (ES) and cumulative abnormal returns (CAR) over a 40-day horizon (CAR_40). Companies with higher earnings surprises demonstrated slightly higher abnormal returns over extended periods; however, the explanatory power of the model was minimal, with an adjusted R-squared value of less than 0.05. This low explanatory power suggests that while there is a relationship, it is not robust enough to provide reliable predictions, highlighting the influence of other variables such as market conditions, investor sentiment, and macroeconomic factors that could better explain stock price movements. The weak relationship indicates that Swedish investors are highly responsive to earnings announcements, leaving little room for prolonged inefficiencies. This aligns with earlier studies, such as Bernard and Thomas (1989), which documented PEAD's diminishing effects in more efficient markets. The lack of a significant relationship over shorter horizons (CAR_5 and CAR_20) suggests that the Swedish stock market processes public earnings information rapidly, reflecting the semi-strong efficiency as proposed by Fama (1970). The swift price adjustments leave limited arbitrage opportunities and indicate the influence of algorithmic trading and institutional investor activities in enhancing market efficiency. The model's minimal explanatory power highlights the complexity of market behavior and suggests that earnings surprises alone are insufficient to predict stock price movements accurately. Market participants consider various factors, including macroeconomic indicators, industry trends, and firm-specific characteristics, which collectively influence stock performance. Additionally, the low R-squared value indicates that much of the variation in cumulative abnormal returns remains unexplained, reinforcing the need for more comprehensive models that integrate qualitative and quantitative factors. 22 The inclusion of control variables, such as momentum and its interaction with ES, did not significantly improve the explanatory power of the model. While ES had a statistically significant effect on CAR_40 in one model, momentum exhibited no meaningful relationship with cumulative abnormal returns. This outcome suggests that price momentum may not play a significant role in the Swedish market compared to earnings surprises, contrasting with findings in less efficient markets where momentum has been shown to have a stronger effect. These findings resonate with Chordia and Shivakumar (2006), who concluded that momentum and earnings surprises independently influence returns, but their combined effects are often subdued in efficient markets. Comparing these findings with other geographical markets, such as the U.S. and European markets, reveals that momentum effects are more pronounced in less efficient markets, where information asymmetry and behavioral biases play a greater role. The long-short portfolio strategy designed to capitalize on PEAD underperformed, with cumulative abnormal returns declining to -8.3% by the end of the observed period. This result underscores the challenge of leveraging PEAD as a viable trading strategy in modern markets. The underperformance is consistent with the findings of Giglio et al. (2008), who noted the erosion of arbitrage opportunities due to growing market efficiency. The results reinforce the idea that trading strategies based on anomalies like PEAD are becoming increasingly less viable in technologically advanced and competitive environments. The inability of the strategy to generate positive abnormal returns suggests that market participants efficiently incorporate earnings information into prices, further supporting the semi-strong form of market efficiency. From a behavioral finance perspective, the weak relationship between ES and CAR_40 suggests that cognitive biases such as overconfidence and herding may have a limited influence on stock price behavior in the Swedish market. This contrasts with Hede (2012), who found more pronounced behavioral effects in less efficient markets, such as emerging economies, where market inefficiencies provide greater opportunities for speculative behavior. Additionally, the results align with Ding et al. (2024), who observed that efficient markets tend to neutralize psychological biases through rapid information dissemination, further emphasizing the reduced significance of behavioral factors in influencing stock price behavior in Sweden. This suggests that Swedish investors may be more rational and informed, reducing the likelihood of behavioral biases distorting market prices. 23 The findings highlight the high level of efficiency in the Swedish stock market, where the rapid adjustment of stock prices to earnings announcements limits the persistence of PEAD. Earlier studies, such as those by Ball and Brown (1968), documented significant drift effects. Still, this study's results suggest that advancements in trading technologies and institutional participation have largely mitigated such anomalies. The limited contribution of control variables, such as momentum, underscores the complexity of stock price behavior and the need for multifaceted approaches incorporating broader macroeconomic and firm-specific factors. Earnings surprises, which are the catalyst for PEAD, occur when a company's reported earnings per share (EPS) deviate from market expectations, typically based on analyst forecasts and market sentiment (Dechow et al., 2010). Positive surprises often lead to price increases, while negative surprises result in declines. However, several factors contribute to the delayed market response observed in PEAD. Behavioral biases, including overconfidence, underreaction, and anchoring, can prevent investors from fully incorporating new information into stock prices, leading to prolonged adjustments (Kahneman & Tversky, 1979; Barberis & Thaler, 2003). Liquidity constraints and limited analyst coverage, particularly in small-cap stocks, can further slow the price adjustment process (Foster et al., 1984). Institutional investors may also contribute to the drift by gradually adjusting their positions to avoid market disruptions. Moreover, the risk-free rate plays a crucial role in evaluating abnormal returns, serving as a benchmark against which investors measure performance. In efficient markets, the expected return of an investment should exceed the risk-free rate only if additional risk is undertaken. However, the presence of PEAD challenges this notion, as it suggests that stock prices may not reflect all available information immediately, allowing for temporary excess returns. This discrepancy underscores the importance of risk assessment when considering investment strategies based on earnings surprises. In a broader geographical context, market efficiency varies significantly across regions. Developed markets, such as those in the U.S. and Western Europe, exhibit higher levels of efficiency due to greater transparency, regulatory oversight, and market liquidity. In contrast, emerging markets often display prolonged inefficiencies driven by lower investor sophistication, limited regulatory frameworks, and information asymmetry. Comparing the 24 Swedish market to these regions highlights its relatively efficient structure but also suggests potential areas where inefficiencies could still arise due to market-specific characteristics. 25 6 Conclusion This study set out to determine whether companies that beat analyst estimates outperform in the stock market, specifically by investigating the persistence of Post-Earnings Announcement Drift (PEAD) in the Swedish market. By analyzing the relationship between earnings surprises (ES) and cumulative abnormal returns (CAR), the findings provide limited evidence to support PEAD. While a weak positive relationship was observed over a 40-day period (CAR_40), the effect was neither strong nor consistent. Over shorter time horizons (CAR_5 and CAR_20), no significant relationship was found, suggesting that the Swedish stock market adjusts quickly and efficiently to new earnings information. These results align with the semi-strong form of the Efficient Market Hypothesis (EMH), which posits that publicly available information is rapidly incorporated into stock prices, leaving limited opportunities for sustained inefficiencies. The analysis of long-short trading strategy based on earnings surprises further reinforces this conclusion. Rather than generating abnormal returns, the strategy consistently underperformed the market, with cumulative abnormal returns declining over the observation period. This outcome highlights the challenges of exploiting anomalies like PEAD in modern markets, where advancements in market technology, the prevalence of algorithmic trading, and high levels of competition have likely diminished arbitrage opportunities. These findings contribute to the broader literature by demonstrating that the theoretical underpinnings of PEAD have limited practical relevance in an efficient market like Sweden. While earlier studies have documented the persistence of PEAD in efficient markets, this research suggests that market dynamics in Sweden are shaped more by efficiency and technological advancements than by behavioral biases or delayed reactions to earnings information. Although this study provides valuable insights, it also raises questions for further research. Future studies could explore additional factors that influence stock price behavior after earnings announcements such as investor sentiment, firm size, or industry-specific characteristics. Moreover, examining the persistence of PEAD in less efficient or emerging markets could provide a richer understanding of how institutional, cultural, and structural factors shape market anomalies. By addressing these areas, future research can continue to refine our understanding of PEAD and its relevance in varying market contexts. 26 References Accounting for Everyone, n.d. Unpacking market earnings: Announcements drive stock price reactions. [online] Available at: https://accountingforeveryone.com/unpacking-market-earnings-announcements-drive-stock-p rice-reactions/ [Accessed 15 January 2025]. American Association of Individual Investors, n.d. Earnings estimates and their impact on stock prices. [online] Available at: https://www.aaii.com/journal/article/earnings-estimates-and-their-impact-on-stock-prices [Accessed 15 January 2025] Barberis, N., & Shleifer, A. (2003). Style investing. Journal of Financial Economics, 68(2), 161–199. Banz, R. W. (1981). The relationship between return and market value of common stocks. Journal of Financial Economics, 9(1), 3-18. Ball, R., & Brown, P. (1968).An empirical evaluation of accounting income numbers. Journal of Accounting Research, 6(2), 159-178. Bernard, V.L., & Thomas, J. K. (1989). Post-earnings-announcement drift: Delayed price response or risk Premium? Journal of Accounting Research, 27, 1-36. Bond, S., Wu, W., & Zheng, S. (2023), 'Seasonal patterns of earnings releases and post-earnings announcement drift', Quantitative Economics, available at: https://ideas.repec.org/a/eee/quaeco/v91y2023icp15-24.html. Chordia, T., & Shivakumar, L. (2006). Earnings and price momentum. Journal of Financial Economics, 80(3), 627–656. Chui, A. C. W., Titman, S. & Wei, K. C. J. (2010). Individualism and Momentum around the World. The Journal of Finance, 65(1), 361-392. Da Silva, S., Matsushita, R., & Giglio, R. (2008). The relative efficiency of stock markets. Economics Bulletin, 7(6), 1–12. Dechow, P.M., Hutton, A.P. and Sloan, R.G., 2010. The relation between analysts' forecasts of long‐term earnings growth and stock price performance following equity offerings. Contemporary Accounting Research, 27(3), pp.637-678. 27 Ding, S., Wang, H., & Sun, Q. (2024). Does continuous good news still mean good news for market volatility? University of Chinese Academy of Social Sciences, Available online 10 December,2024.https://www.sciencedirect.com/science/article/abs/pii/S1544612324016696? getft_integrator=scopus&pes=vor&utm_source=scopus Fama, E. F. (1970). Efficient capital markets: A review of theory and empirical work. The journal of finance, 25(2), 383-417. Fama, E. F. (1991). Efficient Capital Markets: II. Journal of Finance, 46(5), 1575-1617. Hede, P. D. (2012). Behavioral finance. Bookboon.com. ISBN 978-87-403-0200-4. Hong, H., Lim, T., & Stein, J. C. (2000). Bad news travels slowly: Size, analyst coverage, and the profitability of momentum strategies. Journal of Finance, 55(1), 265–295. Hedberg, P. and Lindmark, A., 2012. The Possible Beginning of an End: A Study of the Post Earnings Announcement Drift on the Swedish Stock Market. [online] Available at: https://www.diva-portal.org/smash/get/diva2%3A632109/FULLTEXT01.pdf [Accessed 6 January 2025]. Johansson, P. and Persson, L., 2018. Market efficiency in Nordic stock markets: A post-earnings announcement drift analysis. Nordic Journal of Finance, 12(3), pp.45-67. Kahneman, D., & Tversky, A. (1979). Prospect theory: An analysis of decision under risk. Econometrica, 47(2), 263–291 Livnat, J., & Mendenhall, R. R. (2006). Comparing the post-earnings announcement drift for surprises calculated from analyst and time series forecast. Journal of Accounting Research, 44(1), 177-205 Shleifer, A., 2000. Inefficient markets: An introduction to behavioral finance. Oxford University Press. Wiklund, H. (2020) “An Event Study Examining the Post Earnings Announcement Drift on the Swedish Market”. Lund University Publications. Available at: https://lup.lub.lu.se/student-papers/record/9036272/file/9036275.pdf (Accessed: 5 January 2025). 28