This thesis investigates the price discovery in the foreign exchange market using high frequency data. Traditional exchange rate models assume market homogeneity and the sole existence of public information. However. recent studies suggest such assumptions are not well founded and have generated the 'disconnection' puzzle of exchange rates deviating from their fundamentals in the short and medium term. Using EFX tick-by-tick data, we find that information is not always available to all and the actual price discovery process is dynamic and asymmetric. It suggests that some market participants, trading systems or even exchange rates may possess private information. which helps them to lead others in finding the equilibrium prices. It further reveals the importance of studying the microstructure of the foreign exchange market, which may in the future solve the 'disconnection' puzzle that has baffled the exchange rate theory for the past decades.

This study addresses itself to aspects of the nature of internal auditing. Perhaps we can make this title still more meaningful by focusing on the key terms used. The nature of that is the essential qualities or the general characteristics of something, here it is an activity. 'Internal' - this term is intended to make clear that this is an activity carried on by an organisation itself using its own personnel. Thus, the activity is distinguished from that which is carried out by external public accountants or outside consultants. 'Auditing' - this term suggests a variety of ideas. On the one hand, it can be viewed very narrowly as the mechanical checking of accounts for clerical accuracy and/or on the other hand as a thoughtful investigation and appraisal at the highest operational levels. Definitely this term is intended here to embrace the higher level meaning, even though the lower level activities may also be involved to a certain extent.

Recent research has discussed the possible role of idiosyncratic risk in explaining equity returns. Simultaneously, but somehow independently, numerous other studies have documented the failure ofthe static and conditional capital asset pricing models to explain momentum profits and the value premium The first and second parts ,of this study assess whether the widely documented momentum profits and post-1963 value premium can be attributed to time-varying idiosyncratic risk as described by a GJR-GARCH(l,I)-M model. In accordance with existing studies, we find that the static CAPM has no explanatory power for momentum profits and the value premium, and that firm size has only a limited role to play. The results show that momentum profits are a compensation for time-varying idiosyncratic risk. In addition, negative return shocks. increase the volatility of losers, more than they increase that of winners, and the volatility of the losers responds to news more slowly, but eventually to a greater extent, than that of the winners. The post-1963 value premium can be fully captured' by the conditional variance I specification incorporating time-varying idiosyncratic risk as well. The value premium is a compensation for exposure to ti~e-varying risk. This conclusion is robust to different characteristics of value and growth stocks and to the countries under review (US and UK). The third part ofthis stu~y analyses the impact of trading costs on the profitability of momentum strategies in the UK. It finds that losers are more expensive to trade than winners due to the high selling cost of loser stocks that can be characterized as small size and low trading volume stocks. It proposes a new low-cost momentum strategy by selecting winner and loser stocks with the lowest total transaction costs. While the study severely questions the profitability of standard momentum strategies, it shows that there is still room for momentum-based return enhancement, should asset managers decide to adopt low-cost momentum strategies.

This thesis presents three studies on credit risk modelling. The first study compares the real default probabilities produced by three main structural models of default, Merton model, Longstaff and Schwartz model and Leland and Toft model, to the observed real default probabilities reported by Moody's for the BBB, BB and B rated bonds. We find that none of the models can accurately predict the default probabilities in all these cases. Merton as well as Leland and Toft models underpredict default probabilities. Longstaff and Schwartz model although it produces in some cases Expected Default Frequencies (EDFs) that are close to the observed ones, it tends to overestimate the default probabilities of riskier bonds as well as the default probabilities of bonds with the same rating but higher equity volatility. We also find that structural models tend to underestimate the default probabilities in early years. The second study examines whether information from equity markets, as summarized in the distance to default measure derived from a Merton-Moody's KMV (MKMV) model, provides useful additional information over accounting variables for predicting changes in bank credit ratings. Using a dataset of 98 equity listed banks from 1997 to 2004, we find that di~tance to default measure I has additional explanatory power for modeling current ratings, or predicting credit rating changes over a 6-month or l2-month horizon, but only for the smaller sized banks. We find no evidence that changes in distance-to-default have additional explanatory power for predicting rating categories, regardless of the size ofthe bank. The third study compares two proprietary models, Moody's KMV (MKMV) and BARRA models that use information from the equity and debt market respectively for the estimation of market implied ratings that can be updated continuously. We compare the empirical performance ofthese models in terms of their ability to predict in a timely fashion changes in credit quality by employing a sample of 4594 bonds issued by 447 firms from US for a period of3 years. We find that I?-either model provides a close mapping to observed ratings. Both however are useful for prediction of credit transitions.

Behavioural auditing has been a subject discussed by auditors for many years but little work appears to have been done to quantify fully what it actually is, what its precise definition is and where it fits with other aspects of auditing. This work has, therefore, researched in detail the literature on behavioural auditing. This has encompassed literature written for the three different disciplines within audit - internal audit, external audit and audit consultancy - and has examined the differences between each discipline‟s perceptions of behavioural auditing. The research then considered opinions of behavioural auditing by practioners within each of the disciplines and sought their views on the definition and use of behavioural auditing. By combining the published and practioner views, a definition of behavioural auditing has been postulated. It was then postulated that use of behavioural auditing may make audit work more efficient and effective. This led to a consideration of the interpretation of body language and its potential use by auditors. As with behavioural auditing in general, the use of body language is discussed in audit textbooks, but it appears that little work has been done to establish how it can be used by auditors. Therefore, one aspect of behavioural auditing - the observation of an auditee‟s body language - was tested by inviting students to review five videos which were made to show examples of an auditee‟s differing body language. Feedback about the potential value of behavioural auditing and specifically of the interpretation of body language was gathered and analysed. The results of this feedback have clearly indicated the value of this practical application of behavioural auditing and its use in giving an indication of potential fraud. In addition, the need to provide training to auditors on behavioural auditing and the interpretation body language has been identified.

The first essay, Chapter 3, shows that uninformed investors require a price discount to hold the stock because they perceive a new information uncertainty risk when short-sale constraints are binding and informed trading is absent. Stock prices become less informative when short-sale constraints keep informed investors out of the market. The less informative prices create a new information uncertainty risk for uninformed investors because uninformed investors are unable to figure out the true value of the stock without knowing the private information of informed investors. The new information uncertainty risk effect becomes greater if stocks have greater information uncertainty, which reflects the convenience of learning fundamental news. The second essay, Chapter 4, examines two special new information uncertainty risk effects by controlling for trading volume. When volume is large, uninformed investors observe high buying pressure but cannot distinguish noise demand from information-based buying. They confront this new information uncertainty risk and demand premium to buy stock. Thus, overvaluation caused by short-sale constraints is reduced. When volume is small, uninformed investors convince that informed investors have negative information but do not know how bad the information is. They will not hold the stock under this new information uncertainty risk and therefore future return will becomes worse. The third essay, Chapter 5, studies the impact of informed trading to the momentum effect. It proposes that if momentum is a result of underreaction and if informed trading identifies stocks with underreaction, the presence of informed trading predicts future momentum effect. Consistently, the empirical results show that momentum effect arises when informed trading is present. Greater informed trading leads to greater momentum effect. Although information uncertainty is related to both informed trading and momentum, the identified relationship between informed trading and momentum is robust after controlling for uncertainty.

Using two unique datasets, one at a daily frequency including six currency pairs, and another tick-by-tick dataset in €/US$, we investigate some of the unanswered questions in the field of foreign exchange market microstructure. We confirm the contemporaneous relationship between flows and exchange rates found in the literature in the daily data, but in the forecasting experiments we find no forecasting power, regardless of model, history used forecast horizon or currency pair. The forecasting performance is not improved by considering a system of exchange rates, or by evaluating based on directional ability instead of the more usual RMSE ratio. Subsequently we estimate two standard market microstructure models - Madhavan-Smidt and Huang-Stoll - using the high-frequency dataset in order to gain an insight into the information content of customer order flow. While we are unable to find any evidence of information content from financial customer trades, we find strong evidence that large corporate customer trades are perceived to have statistically and economically significant information content. Lastly we turn our attention to the issue of causality. Using a distributed lag model to investigate the impact of flows on exchange rates and vice versa, corporate orders are found to have a small long-term impact, but more significantly we find evidence of positive feedback trading in both corporate and financial customers. We explore the long-run dynamics of the system using a VECM, and find that all counterparty types have a positive equilibrium relationship with the exchange rate. Crucially, the adjustment dynamics show that all of the weight of adjustment to restore equilibrium after a shock falls to flows. Lastly, we conduct a high frequency forecasting experiment, but again find no evidence of forecasting power. Two important themes emerge from the high-frequency investigation. The first is the apparent importance of corporate customers, and the second is that the direction of causality runs not from flows to exchange rates, but from exchange rates to flows. We conclude that the weight of the evidence suggests that feedback rather than information content is what drives the strong contemporaneous relationship between exchange rates and flows.

Effective financial reporting has become of critical importance in our economic markets and the international accounting scandals of the last decades have accentuated the role of auditing in protecting stakeholders' interests and contributing to an efficient functioning of financial markets. Auditing regulation has been at the centre of recent international debates (e.g. EU Green paper; 2010) and different regulatory interventions have been put in place in different countries and in different periods of time. From a theoretical point of view, there is a broad spectrum of regulatory choices that legislators could take, the extremes being self-regulation and government direct interventions. My empirical works focus on two extreme examples of how regulation may interact with auditing processes with the ultimate objective of improving financial information and therefore enhancing the effectiveness of financial markets. In one case, I analyze whether an example of enforced audit self-regulation is effective in improving audit quality and ultimately, reporting quality. In particular, I analyze the mandatory audit rotation rule, i.e. the rule imposing the change of the auditor after a specified period of time, as a potential means to increase auditor independence and therefore audit and reporting quality. I then study a case of direct monitoring activity on reporting quality carried out by a supervising body, namely the SEC in the U.S., and how this interacts and influences audit processes. The evidence collected may be useful to legislators in order to understand the potential effects of different audit regulatory choices and therefore to effectively address the need of high-quality auditing which strongly characterize our economic markets especially after the financial and accounting scandals.

In this thesis, methods that are capable of improving the revenue and profitability of a financial services company are presented. Of particular interest is the use of customer specific information for pricing insurance products and segmenting a customer population based on the expected profitability of the customers. A prerequisite is the possibility for customers to have many different financial services products from the same provider. The thesis presents multivariate credibility models for how customer specific information from one (or many) financial services products is related to customer specific information from another financial services product. The models are foremost applied to the context of cross-selling (selling additional products to existing customers) where customer specific information from the offered cross-sale product is not available before the sale. As products are related, it is reasonable to use an appropriate (credible) amount of customer specific information from another product (or products), for estimating the profitability expected to emerge from the offered cross-sale product. In four separate but related articles, it is shown that having appropriate models for pricing and customer segmentation is of great importance for a financial services company aiming at running a profitable and growing business.

The long term behaviour of primary product prices has been a central issue underlying projections of commodity price series. Against the background of the Prebisch Singer Hypothesis, the presence, magnitude and direction of a secular trend in commodity price series have themselves become the subject of a long standing debate. This study uses the individual commodity price series underlying the Grilli and Yang data set and, where possible, extends these data series up to 1998. Deflating primary commodity prices by the MUV index, the question of trend components in the time series is studied considering evidence from univariate models and allowing for trend stationary or integrated data series with drift. In this context the impact of serial correlation in finite samples and the impact of wrongly modelling a data series as integrated are considered in detail. Further evidence from a trend test developed by Vogelsang (1998) is also taken into account. In selecting forecast models, the usefulness of unit root pre-testing is assessed allowing for interdependence between the inferred order of integration and the significance of the trend or drift coefficient estimate obtained. Projections from univariate models are obtained for a ten year horizon and Beveridge-Nelson trend cycle decompositions are computed to assess the importance of volatility surrounding the forecasts. It is found that with regards to the past behaviour of primary commodities as well as for the forecasts obtained, the trajectory of primary commodity prices relative to the price of developed country manufactures exports is not generally characterised by a downwards trend.