This thesis investigates systematic liquidity risk and short-term stock price reaction to large one-day price changes. We study 642 constituents of the FTSALL share index over the period from 1st July 1992 to 29th June 2007. We show that the US evidence of a priced systematic liquidity risk of Pastor and Stambaugh (2003) and Liu (2006) is not country-specific. Particularly, systematic liquidity risk is priced in the London Stock Exchange when Amihud's (2002) illiquidity ratio is used as a liquidity proxy. Given the importance of systematic liquidity risk in the asset pricing literature, we are interested in testing whether the different levels of systematic liquidity risk across stocks can explain the anomaly following large one-day price changes. Specifically, we expect that the stocks with high sensitivity to the fluctuations in aggregate market liquidity to be more affected by price shocks. We find that most liquid stocks react efficiently to price shocks, while the reactions of the least liquid stocks support the uncertain information hypothesis. However, we show that time-varying risk is more important than systematic liquidity risk in explaining the price reaction of stocks in different liquidity portfolios. Indeed, the time varying risk explains nearly all of the documented overreaction and underreaction following large one-day price changes. Our evidence suggests that the observed anomalies following large one-day price shocks are caused by the pricing errors arising from the use of static asset pricing models. In particular, the conditional asset pricing model of Harris et al. (2007), which allow both risk and return to vary systematically over time, explain most of the observed anomalies. This evidence supports the Brown et al. (1988) findings that both risk and return increase in a systematic fashion following price shocks. / Yarmouk University, Jordan.

This dissertation consists of three essays, each examining distinct dimensions of cross-sectional variation in exchange rate changes and currency returns conditional on macroeconomic variables. Chapter 2: Protectionism, Bilateral Integration, and the Cross-Section of Ex-change Rate Returns in US Presidential Debates We study the impact of US presidential election TV debates on intraday exchange rates of 96 currencies from 1996 to 2016. Expectations about protectionist measures are the main transmission channel of debate outcomes. Currencies of countries with high levels of bilateral foreign trade with the US depreciate if the election probability of the protectionist candidate increases during the debate. We rationalize our results in a model where a debate victory of a protectionist candidate raises expectations about future tariffs and reduces future net exports to the US, resulting in relative depreciation of currencies with high bilateral trade integration. Chapter 3: Global Portfolio Network and Currency Risk Premia External portfolio investments of countries can explain cross-sectional variation in currency risk premia. Using bilateral portfolio holdings of 26 countries from 2001 to 2021, I construct a network centrality measure where a country is central if it is integrated with key countries that account for a large share in the supply of tradeable financial assets. I find that currency excess returns and interest rates decrease in network centrality. The network centralities are persistent over time and offer a country-specific economic source of risk that are able to explain robust differences in currency risk premia. Empirical asset pricing tests show that the derived risk factor is priced in a cross-section of currency portfolios. Further, negative global shocks cause currencies of central countries to appreciate, while currencies of peripheral countries depreciate. I discuss the findings with implications of a consumption-based capital asset pricing model where central countries have lower consumption growth in high marginal utility states, resulting in an appreciation of their currencies. Chapter 4: FX Dealer Constraints and External Imbalances We study the impact of FX dealer banks' financial health on the cross-sectional variation of exchange rates. Using individual balance sheet information of 39 dealers, we derive an intermediary constraints index that captures the risk-bearing capacity of intermediaries. A deterioration of the solvency of dealer banks impairs their risk-bearing capacity and increases their marginal value of wealth. We test the theoretical prediction of Gabaix and Maggiori (2015) that tightening financial constraints of intermediaries are associated with increasing currency risk premia in the cross-section of the riskiness of currencies, as measured by the net foreign assets of countries. We combine dealer-specific risks to macroeconomic fundamentals of a cross-section of currencies, i.e., the indebtedness to foreigners measured by countries' net foreign assets. We show that currency excess returns increase with a country's external imbalances when constraints are relaxed, but debtor currencies experience a depreciation when constraints tighten.

This project aimed to evaluate the effectiveness of the Auto Regressive Exogenous(ARX) model in forecasting stock prices and contribute to research on statisticalmodels in predicting stock prices. An ARX model is a type of linear regression modelused in time series analysis to forecast future values based on past values and externalinput signals. In this study, the ARX model was used to forecast the closing pricesof stocks listed on the OMX Stockholm 30 (OMXS30*) excluding Essity, Evolution,and Sinch, using historical data from 2016-01-01 to 2020-01-01 obtained from YahooFinance. The model was trained using the least squares approach with a control signal that filtersoutliers in the data. This was done by modeling the ARX model using optimizationtheory and then solving that optimization problem using Gurobi OptimizationSoftware. Subsequently, the accuracy of the model was tested by predicting prices in aperiod based on past values and the exogenous input variable. The results indicated that the ARX model was not suitable for predicting stock priceswhile considering short time periods.

No description available.

No description available.

Syfte: Syftet är att testa Fama och Frenchs femfaktormodell på den svenska aktiemarknaden. Detta genom att undersöka huruvida modellen kan statistiskt förklara portföljers genomsnittliga avkastning samt ifall specifika faktorer har statistisk signifikans. Metod: En kvantitativ studie med ett deduktivt förhållningssätt. Undersökningen utför tester på den svenska aktiemarknaden mellan 2015-01-01 och 2019-12-31 genom en regressionsanalys. Upptäckter: Fama och Frenchs femfaktormodell förkastas som helhet men det påvisas däremot att HML är statistisk signifikant inom sex av sex storlekssorterade portföljer, följt av SMB med fyra av sex. Fama och Frenchs femfaktormodellen har svårigheter att förklara avkastningen för mindre företag sorterade utifrån lönsamhet och book-to-market tal. Forskningsimplikationer: Undersöker ett forskningsämne som eftersträvar studier och tester på ett flertal varierande marknader för att förklara aktiers avkastningsmönster. Orginalitet och värde: Studien särskiljer sig på grund av avsaknaden av forskning på den svenska aktiemarknaden. Därtill bidrar studien till ett undersökningsområde för små öppna ekonomier som den svenska marknaden grundas i. / Purpose: The purpose is to test Fama and French's five-factor model in the Swedish stock market. This is done by examining whether the model can explain portfolios' average return and whether specific factors have statistical significance. Method: A quantitative study with a deductive approach. The survey performs tests on the Swedish stock market between 2015-01-01 and 2019-12-31 through a regression analysis. Findings: Fama and French's five-factor model is rejected as a whole, but it is shown that HML is statistically significant in every size-sorted portfolio, followed by SMB with statistical significance in four out of six portfolios. Fama and French's five-factor model have difficulty explaining the returns for smaller companies sorted on profitability and book-to-market ratio. Research implications: Investigates a research topic that strives for an increased number of studies and tests in different markets to explain stock return patterns. Originality and value: The study differs due to the lack of research on the Swedish stock market. In addition, the study contributes to a study area for small open economies in which the Swedish market is based.

In recent years, machine learning models have gained traction in the field of empirical asset pricing for their risk premium prediction performance. In this thesis, we build upon the work of [1] by first evaluating models similar to their best performing model in a similar fashion, by using the same dataset and measures, and then expanding upon that. We explore the impact of different feature extraction techniques, ranging from simply removing added complex- ity to representation learning techniques such as incremental PCA and autoen- coders. Furthermore, we also introduce recurrent connections with LSTM and combine them with the earlier mentioned representation learning techniques. We significantly outperform [1] in terms of monthly out-of-sample R2, reach- ing a score of over 3%, by using a condensed version of the dataset, without interaction terms and dummy variables, with a feedforward neural network. However, across the board, all of our models fall short in terms of Sharpe ratio. Even though we find that LSTM works better than the benchmark, it does not outperform the feedforward network using the condensed dataset. We reason that this is because the features already contain a lot of temporal information, such as recent price trends. Overall, the autoencoder based models perform poorly. While the linear incremental PCA based models perform better than the nonlinear autoencoder based ones, they still perform worse than the bench- mark. / Under de senaste åren har maskininlärningsmodeller vunnit kredibilitet inom området empirisk tillgångsvärdering för deras förmåga att förutsäga riskpre- mier. I den här uppsatsen bygger vi på [1]s arbetet genom att först implemente- ra modeller som liknar deras bäst presterande modell och utvärdera dem på ett liknande sätt, genom att använda samma data och mått, och sedan bygga vida- re på det. Vi utforskar effekterna av olika variabelextraktionstekniker, allt från att helt enkelt ta bort extra komplexitet till representationsinlärningstekniker som inkrementell PCA och autoencoders. Vidare introducerar vi även LSTM och kombinerar dem med de tidigare nämnda representationsinlärningstekni- kerna. Min bästa modell presterar betydligt bättre än [1]s i termer av månatlig R2 för testdatan, och når ett resultat på över 3%, genom att använda en kompri- merad version av datan, utan interaktionstermer och dummyvariabler, med ett feedforward neuralt nätverk. Men överlag så brister alla mina modeller i ter- mer av Sharpe ratio. Även om LSTM fungerar bättre än riktvärdet, överträffar det inte feedforward-nätverket med den komprimerade datamängden. Vi re- sonerar att detta är på grund av inputvariablerna som redan innehåller en hel del information över tid, som de senaste pristrenderna. Sammantaget presterar de autoencoderbaserade modellerna dåligt. Även om de linjära inkrementell PCA-baserade modellerna presterar bättre än de olinjära autoencoderbaserade modellerna, presterar de fortfarande sämre än riktvärdet.

This thesis investigates the complex dynamics of foreign exchange (FX) risk affecting Swedish multinational corporations and their financial performance, with a focus on the impact of company size and periods of economic crisis. Amidst global economic interdependencies, these entities encounter substantial FX risks, primarily due to the volatility of the Swedish Krona (SEK) against major currencies. Utilizing a comprehensive dataset spanning from 2004 to 2023, this study employs an empirical approach grounded in the International Capital Asset Pricing Model (ICAPM) and Purchasing Power Parity (PPP) to analyze the correlation between currency fluctuations and stock valuations. The analysis reveals that both company size and economic crises significantly modulate the effects of FX risks, with larger companies often better positioned to manage these risks through sophisticated hedging strategies. Smaller firms, conversely, show greater sensitivity to economic disruptions, particularly during crises which heighten the volatility of FX impacts. The findings indicate that FX risks significantly influence the financial outcomes of these firms, with both direct impacts on stock returns and indirect effects through operational strategies. The thesis underscores the importance of robust risk management strategies and the potential for policy adjustments to mitigate adverse effects from currency volatility. The insights derived from this research aims to contribute to a deeper understanding of the financial economics of foreign exchange, providing implications for investors and multinational corporations operating in global markets.

<p>The cross-sectional variation of stock returns used to be described by the Capital Asset Pricing Model until the early 90‟s. Anomalies, such as, book-to-market effect and small firm effect undermined CAPM‟s ability to explain stock returns and Fama & French (1992) have shown that simple firm attributes, like, firm size and book-to-market value can explain the returns far better than Beta. Following Fama & French many other researchers examine the explanatory powers of CAPM and other asset pricing models. However, most of those studies use US data. There are some researches done in different countries than US, however more out-of-sample studies need to be conducted.</p><p>To our knowledge there are very few studies using the Swedish data and this thesis contributes to that small pool of studies. Moreover, the studies testing the CAPM use the unconditional version of the model. There are some papers suggesting the use of a conditional CAPM that would exhibit better explanatory powers than the unconditional CAPM. Different ways of conditioning the CAPM have been proposed, but one that we think is the least complex and possible to make use of in the business world is the dual-beta model. This conditional CAPM assumes a different relationship between beta and stock returns during the up markets and down markets. Furthermore, the model has not thoroughly been tested outside the US. Our study is the first to use the dual-beta model in Sweden. In addition, the momentum effect has lately been given some attention and Fama & French‟s (1993) three factor model has not been able to explain the abnormal returns related to that anomaly. We test the Fama & French three factor model, CAPM and Carhart‟s four factor model‟s explanatory abilities of the momentum effect using Swedish stock returns. Ultimately, our aim is to find the best model that describes stock return cross-section on the Stockholm Stock Exchange.</p><p>We use returns of all the non-financial firms listed on Stockholm Stock Exchange between September, 1997 and April, 2010. The number of companies included in our time sample is 366. The results of our tests indicate that the small firm effect, book-to-market effect and the momentum effect are not present on the Stockholm Stock Exchange. Consequently, the CAPM emerges as the one model that explains stock return cross-section better than the other models suggesting that Beta is still a proper measure of risk. Furthermore, the conditional version of CAPM describes the stock return variation far better than the unconditional CAPM. This implies using different Betas to estimate risk during up market conditions and down market conditions.</p>

The cross-sectional variation of stock returns used to be described by the Capital Asset Pricing Model until the early 90‟s. Anomalies, such as, book-to-market effect and small firm effect undermined CAPM‟s ability to explain stock returns and Fama &amp; French (1992) have shown that simple firm attributes, like, firm size and book-to-market value can explain the returns far better than Beta. Following Fama &amp; French many other researchers examine the explanatory powers of CAPM and other asset pricing models. However, most of those studies use US data. There are some researches done in different countries than US, however more out-of-sample studies need to be conducted. To our knowledge there are very few studies using the Swedish data and this thesis contributes to that small pool of studies. Moreover, the studies testing the CAPM use the unconditional version of the model. There are some papers suggesting the use of a conditional CAPM that would exhibit better explanatory powers than the unconditional CAPM. Different ways of conditioning the CAPM have been proposed, but one that we think is the least complex and possible to make use of in the business world is the dual-beta model. This conditional CAPM assumes a different relationship between beta and stock returns during the up markets and down markets. Furthermore, the model has not thoroughly been tested outside the US. Our study is the first to use the dual-beta model in Sweden. In addition, the momentum effect has lately been given some attention and Fama &amp; French‟s (1993) three factor model has not been able to explain the abnormal returns related to that anomaly. We test the Fama &amp; French three factor model, CAPM and Carhart‟s four factor model‟s explanatory abilities of the momentum effect using Swedish stock returns. Ultimately, our aim is to find the best model that describes stock return cross-section on the Stockholm Stock Exchange. We use returns of all the non-financial firms listed on Stockholm Stock Exchange between September, 1997 and April, 2010. The number of companies included in our time sample is 366. The results of our tests indicate that the small firm effect, book-to-market effect and the momentum effect are not present on the Stockholm Stock Exchange. Consequently, the CAPM emerges as the one model that explains stock return cross-section better than the other models suggesting that Beta is still a proper measure of risk. Furthermore, the conditional version of CAPM describes the stock return variation far better than the unconditional CAPM. This implies using different Betas to estimate risk during up market conditions and down market conditions.