The evidence presented in this study suggests that the dispersion, accuracy and transitory component in revisions of financial analysts' forecasts (FAF) are determined by production/investment/financing decisions, accounting choices as well as firm specific characteristics including the type of control, debt to equity ratio and size of the firm. Firms with managers control (owners control), high (low) debt to equity ratio and large (small) size are hypothesized to have higher (lower) dispersion, forecast error and transitory components in revisions of FAF. These hypotheses are motivated by the contracting cost and political visibility theories. The information availability theory is included as a contrast to the political visibility hypothesis. The information availability hypothesis predicts large (small) firms to have lower (higher) dispersion, forecast error and transitory component in revisions of FAF. The regression results are sensitive to deflated and undeflated measures of the dispersion and accuracy of FAF and size of the firm. The appropriateness of the two measures of firm's size, the book value of total assets and the market value of common stock plus long-term debt, as well as the deflated and undeflated measures of dispersion and accuracy of FAF are investigated. It is concluded that deflated measures of the dispersion and forecast errors and the market value as measure of firm size are misspecified in the present context. The current year forecast revisions are assumed to consist of the transitory and permanent components. The second year forecast revisions are used to represent the long-term forecast revisions and are used as a control for the permanent component of forecast revisions. The regression results are consistent with the contracting and political visibility hypotheses. The firm specific characteristics are hypothesized to influence forecast errors and dispersion directly and indirectly through business risk and accounting policy choices. The links between firm characteristics and business risk, accounting policy choices, dispersion and forecast errors are established and path analysis is used to test these relationships. These relationships are observed to be consistent with predictions and significant.

No description available.

This thesis analyzes estimation methods and testing procedures for handling large data series. The first chapter introduces the use of the adaptive elastic net, and the penalized regression methods nested within it, for estimating sparse vector autoregressions. That chapter shows that under suitable conditions on the data generating process this estimation method satisfies an oracle property. Furthermore, it is shown that the bootstrap can be used to accurately conduct inference on the estimated parameters. These properties are used to show that structural VAR analysis can also be validly conducted, allowing for accurate measures of policy response. The strength of these estimation methods is demonstrated in a numerical study and on U.S. macroeconomic data. The second chapter continues in a similar vein, using the elastic net to estimate sparse vector autoregressions of realized variances to construct volatility forecasts. It is shown that the use of volatility spillovers estimated by the elastic net delivers substantial improvements in forecast ability, and can be used to indicate systemic risk among a group of assets. The model is estimated on realized variances of equities of U.S. financial institutions, where it is shown that the estimated parameters translate into two novel indicators of systemic risk. The third chapter discusses the use of the bootstrap as an alternative to asymptotic Wald-type tests. It is shown that the bootstrap is particularly useful in situations with many restrictions, such as tests of equal conditional predictive ability that make use of many orthogonal variables, or `test functions'. The testing procedure is analyzed in a Monte Carlo study and is used to test the relevance of real variables in forecasting U.S. inflation.

Detection of 'anomalies', empirical regularities that are inexplicable within a preeminent or accepted paradigm, is a key aspect of the operation of scientific endeavour. The dominant theories of financial economics, those deriving from the CAPM/APT literature, hold that there should not exist persistent differences in the returns to assets across calendar frequencies. An extensive review of the literature reveals that in a wide variety of assets and markets there is evidence that returns differ according to the calendar frequency, in particular across days of the week and months of the year and around recurrent holidays. However, this review also reveals considerable room for increased methodological and statistical sophistication. In particular, the nature and extent of the data indicate that techniques based on robust regression, non-parametric statistics and Bayesian inference are more appropriate than the predominantly OLS based approaches displayed in the literature. Papers that adopt these more sophisticated approaches generally find much weaker evidence for such calendar anomalies. In essence, the Irish Stock Exchange operated free from exchange controls and in a broadly homogenous monetary and economic environment from 1988 to 1998. Daily returns from 1988 to 1998, on official equity indices, and from 1993 to 1998 on equal and value weighted equity indices, are examined. The evidence is that even when more sophisticated and appropriate techniques are used there is still some evidence for a daily pattern in the returns to these indices. However this pattern is dissimilar to that found elsewhere, consisting of a midweek positive peak as opposed to the more commonly found low returns at the start of the week and higher returns on Friday. This pattern is not a function of the settlement system, does not appear to be related to the pattern of either microeconomic (firm-specific) or macroeconomic information releases, nor does it appear to be a function of endogenous news generation. Previous international research indicates a January peak in returns, while previous research on the Irish market had also found an April peak. While the investigation here of the monthly pattern of returns confirms, in a statistically and methodologically robust manner, the January peak no evidence is found of an April peak. Examination of the return pattern around exchange holidays indicates that, in common with other markets referenced in the literature, there is a rise in returns before a holiday. However, on decomposition into local and international components we find that although the local effect is strong this effect is negative, which is a major point of departure from previous research findings.

M.Com. (Financial Economics) / Forecasting inflation is an important concern for economists and business alike throughout the world. Despite the relative success of macroeconomic forecasting models in forecasting inflation, there is potential to improve these models to account for nonlinear relationships between inflation and the chosen independent variables. Artificial neural networks (ANNs) have found increased applicability as a potential nonlinear forecasting tool that accounts for nonlinearity found in data. In this study, we investigate the ability of genetically optimised neural networks to forecast South African inflation. The results were compared to economic forecasts obtained from traditional econometric models as well as macroeconomic structural models. The results obtained show that the genetically optimised neural networks indicate some ability to be used as potential forecasting tools. Their biggest advantage over the traditional forecasting techniques is that they do not impose the restriction of linearity on the data to be forecasted.

Not available / text

A thesis submitted to the Faculty of Commerce, Law and Management, University of the Witwatersrand, Johannesburg, in fullment of the requirements for the Doctor of Philosophy in Economics Degree, June 2017 / This thesis contains three main chapters. The rst chapter employs wageprice spirals to generate ination forecasts for Australia, Canada, France, South Korea, South Africa, United Kingdom and the United States. We use three competing specications of the wage-price spirals, and test which specication provides the best forecasts of price ination. For each specication we provide one quarter, four quarter and eight quarter ahead dynamic forecasts of price ination. The rst two wage-price spirals in the rst chapter are from the Keynesian tradition from te standpoint of expectations formation. The chapter also considers the New Keynesian wage-price spiral. We use the Root Means Square Error and the Clark and West statistic to compare the performance of ination forecasts from the three competing wage-price spirals that we consider in the rst chapter of the thesis. We nd that the New Keynesian wage and price specication su⁄ers from the wrong sign problem, and its forecasts of price ination generally outperform those from the old Keynesian wage price spiral for the eight quarter ahead time horizon. The usefulness of this nding to the conduct of monetary policy is limited due to the wrong sign problem of the forcing variable in the New Keynesian wageprice spiral. We also nd that the Flaschel type specication of price and wage ination produce four and eight quarter head ination forecasts that are better than those from the Fair type specication. We further nd that the Fair type specication price and wage equation produce the best forecasts of ination for the one quarter ahead time horizon. In the second chapter, we estimate natural variables and test their ability to explain the ination process for the eight countries that we consider. We use the traditional Keynesian wage-price spiral and the triangle system approaches to estimate the NAIRU and potential output. In the case of the traditional Keynesian wage-price spiral, the price Phillips curve, which can be specied as a triangle Phillips curve, features backward looking ination expectations and nominal wage ination, the output gap and supply shocks. The nominal wage Phillips curve features ination expectations and price ination and the unemployment gap. The presence of price ination in the nominal wage Phillips curve and the presence of nominal wage ination in the price Phillips curve leads to the interaction between the two Phillips curves. The separate demand pressure terms allows for their identication since, as someauthorsintheliteraturearguethatthegoodsandlabourmarketsdonot move in line with each other. To compute the NAIRU and potential output using the Keynesian approach, we rstly exploit the information contained in vector of unobservable by estimating the wage-price spiral in di⁄erence form using the Seemingly Unrelated Regression method. We use this regression method in order to control for any correlation that may exist between errors in the price and wage Phillips curves. This allows us to solve for the vector of potential output and the NAIRU. We then the moving average technique in order to avoid problems associated with the HP lter for smoothing. Due to data availability, use the MA (20) approximation of the low pass lter after padding the endpoints with forecasts from an AR(4) process. We follow a similar procedure in the estimation of the estimation of the NAIRU and potential output for the triangle system approach. To test which method produces the best natural variables, we t the gaps that are computed from the NAIRU and potential output in a simple single equation price Phillips curve. To test which specication produces the best natural varibles we use a simple single equation triangle price Phillips curve. We nd that the output gaps computed from the two competing approaches are signicantly correlated, the same applies to the unemployment gaps computed from the two approaches. We nd that the quality of unemployment rate gaps computed from the Keynesian and triangle system approach to produce similar quality of results when tted to a single equation triangle price Phillips curve. The Keynesian approach slightly outperforms the triangle systems approach in the when considering the output gap as a proxy for the demand pressure. These results indicate that the wage-price spiral still remains an important tool in the determination of the dynamics ination. In the third chapter, we analyze the relationship between monetary policy and natural variables for Australia, Canada, France, South Korea, South Africa, United Kingdom and the United States. We do this by specifying a relationship between natural rates and the real interest rate. The theoretical relationship between the two variables is positive in the case of the NAIRU and negative through Okuns law in the case of potential output. We regress the natural variable against a constant and the MA(8) of the real interest rate. We nd that the parameter of the real interest rate generally has a correct sign when considering the Keynesian approach computed NAIRUs, with only four being signicant. In the case of the triangle system approach NAIRU, we nd that the real interest rate parameter has a correct sign and signicant four countries. We nd that NAIRUs computed using di⁄erent methodologies can produce a di⁄erent reference point for policy makers. We then introduce hysteresis in the relationship between monetary policy and the NAIRU. We then nd that the interest rate parameter generally has a incorrect sign across the three approaches. The HP ltering approach which we include in our study for comparison purposes produces incorrect correlation for all the countries, while the Keynesian approach negative correlation for seven countries, and the triangle system approach in six countries. In the case of the relationship between monetary policy and potential output, we nd that the real interest rate parameter has an incorrect sign. When introducing hysteresis in the relationship between monetary policy and potential we nd that, unlike in the case of the NAIRU this plays signicant role in the relationship. / XL2018

In the first chapter, I investigate how external analyst forecasts influence managerial earnings decisions. Using shifts in analyst composition effected by brokerage mergers as a source of exogenous variation, I establish a one-to-one response of firm earnings to analyst forecasts. This response is driven by accounting accruals, consistent with short-termist earnings management. I find that the market perceives these accruals as costly to the firm. I present a model where this behavior emerges as a rational equilibrium, confirmed by a calibration that mirrors a one-to-one forecast-earnings relationship. Calibration outcomes align with real-world earnings and forecast patterns. In the second chapter (co-authored with Harrison Hong and Jeffrey Kubik), we estimate the cost to capital of climate policy. Many US states have set ambitious renewable portfolio standards (RPS) that require utilities to switch from fossil fuels toward renewables. RPS increases the renewables capacity, bond issuance, maturity, and yield spreads of investor-owned utilities compared to municipal producers that are exempted from this climate policy. Contrary to stranded-asset concerns, the hit to overall firm financial health is moderate. Falling cost of renewables and pass through of these costs to consumers mitigate the burden of RPS on firms. Using a Tobin’s 𝒒 model, we show that, absent these mitigating factors, the impact of RPS on firm valuations would have been severe. In the third chapter (co-authored with Lukas Fischer), we identify a source of peer group influence that is plausibly orthogonal to information provision, yet nonetheless affects economic decision-making: the shock to an equity analyst of their undergraduate college football team winning the NCAA Championship Game. We find that analysts’ forecasts respond positively to their undergraduate school’s football team winning the NCAA final. We then show that the shock of ‘winning’ spreads within an analyst’s brokerage, positively influencing the forecasts of their colleagues. Brokerages where the degree of this diffusion is greater have lower female representation in their analyst teams, as well as lower ESG scores.

每股盈餘是公司的重要財務資訊之一，它可以反應公司的經營績效，因此一方面可以提供給投資者作為投資決策之參考，另一方面提供給管理者作為管理評量的參考指標之一。過去在每股盈餘等財務預測往往以統計方法進行，因此在自變數選擇上常受到限制，同時有些預測模式其輸出結果往往只能以常長或衰退等二元式的結果表示。而另一方面，以類神經網路預測方式的預測模式可能因變數增加，使得網路變的較複雜。本研究嘗試以人工智慧中的遺傳演算法來作為預測的工具，發展財務預測模型，來預測每股盈餘，解決過去預測方式的限制或缺點。同時也將對過去的遺傳演算法稍做修正，並嘗試以實際值的編碼方式進行編碼，以符合需求。最後進一步比較遺傳演算法和其他預測方式，瞭解以遺傳演算法做於預測每股盈餘工具的特性及優缺點。 / Earnings per share (EPS) is one of the important financial indicators to a corporation. It reflects the operating performance of a corporation. On one hand, EPS provides information available to investors for decision making; on the other hand, it is an indexfor measurement of management. In the past, financial forecasting was often done by using statistical models. However, the input variables were limited by using these statistical models. Besides, some stastical models only provide dichotomy output ,such as either "grwoth" or"decline". The neural network forecasting model will be more of complexity, when the input variable increases. This research attempts to develop a financial forecasting model to forecast the EPS by using the Genetic Algorithms, which is a new topic of artificial intelligence. This model excludes both the limitations and disadvantages of the models mentioned above. Here, the genetic algorithms will be modified and the real number will be used to code as a gene of achromosome to meet the requirements of the finacial model. Finally,we compare the genetic algorithms financial forecasting model with the other ones in order to understand the features, advantages and disadvantages of genetic algorithms as being a financial forecasting tool .