[pt] Este estudo explora fontes de dados não convencionais como dados textuais de jornais e pesquisas de internet do Google Trends em dois problemas empíricos: (i) analisar o impacto da mobilidade sobre o número de casos e mortes por Covid-19; (ii) nowcasting do PIB em alta-frequência. O primeiro artigo usa fontes de dados não estruturados como controle para fatores comportamentais não observados e encontra que um aumento na mobilidade residencial diminui significativamente o número de casos e mortes num horizonte de quatro semanas. O segundo artigo usa fontes de dados não estruturadas para fazer um nowcasting semanal do PIB, mostrando que dados textuais e Google Trends pode aumentar a qualidade das projeções (medido pelo EQM, EAM e outras métricas) comparado com as expectativas de mercado do Focus como base. Em ambos casos, dados não estruturados reveleram-se fontes ricas de informação não codificadas em indicadores estruturados convencionais. / [en] This study exploits non-conventional data sources such as newspaper textual data and internet searches from Google Trends in two empirical problems: (i) analysing the impacts of mobility on cases and deaths due to Covid-19; (ii) nowcasting GDP in high-frequency. The first paper resorts to unstructured data to control for non-observable behavioural effects and finds that an increase in residential mobility significantly reduces Covid-19 cases and deaths over a 4-week horizon. The second paper uses unstructured data sources to nowcast GDP on a weekly basis, showing that textual data and Google Trends can significantly enhance the quality of nowcasts (measured by MSE, MAE and other metrics) compared to Focus s market expectations as a benchmark. In both cases, unstructured data was revealed to be a valuable source of information not encoded in structured indicators.

Fear of crime may differ between areas, even if levels of crime do not differ. Policymakers and companies should be interested in how much economical values individuals put on their emotions. No previous paper has tried to estimate the economic consequences of the kind of fear of crime that does not stem from an increase in crime. Through a hedonic fixed effect-approach and a unique data set, I close this gap by estimating the willingness to pay to avoid that fear. As the outcome variable, I use sold apartments in Stockholm municipality in the years 2017 and 2020. I find suggestive evidence of small to moderate effect of fear of crime on housing prices, even after controlling for crime levels, with an elasticity of -2% to -6%. The results are robust throughout different robustness tests. These findings may help politicians in their cost-benefit analyses when planning safety-increasing projects.

In this study, a panel data set of flights made by employees at the Royal Institute of Technology (KTH) in Sweden is analyzed using generalized linear modeling approaches, with the aim to create a model with high predictive capability of the quarterly CO2 emission and the number of flights, for a year not included in the model estimation. A Zero-inflated Gamma regression model is fitted to the CO2 emission variable and a Zero-inflated Negative Binomial regression model is used for the number of flights. To build the models, cross-validation is performed with the observations from 2018 as the training set and the observations from the next year, 2019, as the test set. One at a time, the variable that best improves the prediction of the test set data (either as included in the count model or the zero-inflation model) is selected until an additional variable turns out insignificant on a 5% significance level in the estimated model. In addition to the variables in the data, three lags of the dependent variables (CO2 emission and flights) were included, as well as transformed versions of the continuous variables, and a random intercept each for the categorical variables indicating quarter and department at KTH, respectively. Neither model selected through the cross-validation process turned out to be particularly good at predicting the values for the upcoming year, but a number of variables were proven to have a statistically significant association with the respective dependent variable.

This study examines the fundamental determinants of bilateral trade flows between South Africa and BRIC countries. This is done by exploring the magnitude of exports among these countries. The Gravity model approach is used as the preferred theoretical framework in explaining and evaluating successfully the bilateral trade flows between South Africa and BRIC countries The empirical part of this study uses panel data methodology covering the time period 2000-2012 and incorporates the five BRICS economies in the sample. The results of the regressions are subject to panel diagnostic test procedures. The study reveals that, on the one hand, there are positive and significant relationships between South African export flows with the BRICs and distance, language dummy, the BRICs’ GDP, the BRICs’ openness and population in South Africa. On the other hand, GDP in South Africa, real exchange rate and time dummy are found to be negatively related to export flows. / Economics / M. Com. (Economics)

本論文的研究主軸圍繞於自我迴歸模型的估計與推論上。文獻上自我迴歸模型的估計多直接採用最小平方法, 但此估計方式卻有兩個缺點:(一)當序列具單根時,最小平方估計式的漸近分配為非正規型態,因此檢定時需透過電腦模擬得到臨界值;(二)最小平方估計式雖具一致性,但卻有嚴重的有限樣本偏誤問題。有鑑於此,我們提出一種「二階差分轉換估計式」,並證明該估計式的偏誤遠低於前述最小平方估計式,且在序列為粧定與具單根的環境下具有相同的漸近常態分配。此外,二階差分轉換估計式相當適合應用於固定效果追蹤資料模型,而據以形成的追蹤資料單根檢定在序列較短的情況下仍有不錯的檢定力。 本論文共分四章,茲分別簡單說明如下: 第1章為緒論,回顧文獻上估計與推論自我回歸模型時的問題,並說明本論文的研究目標。估計自我迴歸模型的傳統方式是直接採取最小平方法,但在序列具單根的情況下由於訊息不隨時間消逝而快速累積,使估計式的收斂速度高於序列為恒定的情況。不過,這也導致最小平方估計式的漸近分配為非標準型態,並使得進行假設檢定前必須先透過電腦模擬來獲得臨界值。其次,最小平方估計式雖具一致性,但在有限樣本下卻是偏誤的。實證上, 樣本點不多是研究者時常面臨的窘境,並使得小樣本偏誤程度格外嚴重。本章中透過對前述問題形成因素的瞭解,說明解決與改善的方法,亦即我們提出的「二階差分轉換估計式」。 第2章主要目的在於推導二階差分轉換估計式之有限樣本偏誤。我們亦推導了多階差分自我迴歸模型下二階段最小平方估計式(two stage least squares, 2SLS)與 Phillips andHan (2008)採用的一階差分轉換估計式之偏誤,以同時進行比較。本章理論與模擬結果皆顯示,一階與二階差分轉換估許式與2SLS之 $T^{−1}$ 階偏誤程度皆低於以最小平方法估計原始準模型(level model)的偏誤,其中 T 為時間序列長度。另外,一階差分轉換估計式與二階差分轉換估計式在 $T^{−1}$ 階偏誤上,分別與一階和二階差分模型下2SLS相同,但兩估計式的相對偏誤程度則因自我相關係數的大小而互有優劣。同時,我們發現估計高於二階的差分模型對小樣本偏誤並無法有更進一步的改善。最後,即使在樣本點不多的情況下,本章所推導的偏誤理論對於實際偏誤仍有良好的近似能力。 第3章主要目的在於發展二階差分轉換估計式之漸近理論。與 Phillips and Han (2008) 採用之一階差分轉換估計式相似的是,該估計式在序列為恒定與具單根的情況下收斂速度相同,並有漸近常態分配的優點。值得注意的是, 二階差分轉換估計式的漸近分配為 N(0,2),不受任何未知參數的影響。另外,當序列呈現正自我相關時,二階差分轉換估計式相較於一階差分轉換估計式具有較小的漸近變異數,進而使得據以形成的檢定統計量有較佳的對立假設偵測能力。最後, 誠如 Phillips and Han (2008) 所述,由於差分過程消除了模型中的截距項,使得此類估計方法在固定效果的動態追蹤資料模型(dynamic panel data model with fixed effect) 具相當的發展與應用價值。 本論文第4 章進一步將二階差分轉換估計式推展至固定效果的動態追蹤資料模型。文獻上估計此種模型通常利用差分來消除固定效果後,再以一般動差法 (generalized method of moments, GMM) 進行估計。然而,這樣的估計方式在序列為近單根或具單根時卻面臨了弱工具變數(weak instrument)的問題,並導致嚴重的估計偏誤。相反的,差分轉換估計式所利用的動差條件在近單根與單根的情況下仍然穩固,因此在小樣本下的估計偏誤相當輕微(甚至無偏誤)。另外,我們證明了不論序列長度(T )或橫斷面規模(n)趨近無窮大,差分轉換估計式皆有漸近常態分配的性質。與單一序列時相同的是,我們提出的二階差分轉換估計式在序列具正自我相關性時的漸近變異數較一階差分轉換估計式小;受惠於此,利用二階差分轉換估計式所建構的檢定具有較佳的檢力。值得注意的是,由於二階差分轉換估計式在單根的情況下仍有漸近常態分配的性質,我們得以直接利用該漸近理論建構追蹤資料單根檢定。電腦模擬結果發現,在小 T 大 n 的情況下,其檢力優於文獻上常用的 IPS 檢定(Im et al., 1997, 2003)。 / This thesis deals with estimation and inference in autoregressive models. Conventionally, the autoregressive models estimated by the least squares (LS) procedure may be subject to two shortcomings. First, the asymptotic distribution of the LS estimates for autoregressive coefficient is discontinuous at unity. Test statistics based on the LS estimates thus follow nonstandard distributions, and the critical values obtained need to rely on Monte Carlo techniques. Secondly, as is well known, the LS estimates of autoregressive models are biased in finite samples. This bias could be substantial and leads to serious size distortion for the test statistics built on the estimates and forecast errors. In this thesis,we consider a simple newmethod ofmoments estimator, termed the “transformed second-difference” (hereafter TSD) estimator, that is without the aforementioned problems, and has many useful applications. Notably, when applied to dynamic panel models, the associated panel unit root tests shares a great power advantage over the existing ones, for the cases with very short time span. The thesis consists of 4 chapters, which are briefly described as follows. 1. Introduction: Overview and Purpose This chapter first reviews the literature and states the purpose of this dissertation. We discuss the sources of problems in estimating autoregressive models with the conventional method. The motivation to estimate the autoregressive series with multiple-difference models, instead of the conventional level model, is provided. We then propose a new estimator, the TSD estimator, which can avoid (fully or partly) the drawbacks of the LS method, and highlight its finite-sample and asymptotic properties. 2. The Bias of 2SLSs and transformed difference estimators in Multiple-Difference AR(1) Models In this chapter, we derive approximate bias for the TSD estimator. For comparisons, the corresponding bias of the two stage least squares estimators (2SLS) in multiple-difference AR(1) models and the transformed first-difference (TFD) estimator proposed by Chowdhurry (1987) are also given as by-products. We find that: (i) All the estimators considered are much less biased than the LS ones with the level regression; (ii)The difference method can be exploited to reduce the bias only up to the order of difference 2; and (iii) The bias of the TFD and TSD estimators share the same order at $O(T^{-1})$ as that of 2SLSs. However, to the extent of bias reductions, neither the 2 considered transformed difference estimators shows a uniform dominance over the entire parameter space. Our simulation evidence lends credible supports to our bias approximation theory. 3. Gaussian Inference in AR(1) Time Series with or without a Unit Root The goal of the chapter is to develop an asymptotic theory of the TSD estimator. Similar to that of the TFD estimator shown by Phillips and Han (2008), the TSDestimator is found to have Gaussian asymptotics for all values of ρ ∈ (−1, 1] with $\sqrt{T}$ rate of convergence, where ρ is the autoregressive coefficient of interest and T is the time span. Specifically, the limit distribution of the TSD estimator is N(0,2) for all possible values of ρ. In addition, the asymptotic variance of the TSD estimator is smaller than that of the TFD estimator for the cases with ρ > 0, and the corresponding t -test thus exhibits superior power to the TFD-based one. 4. Estimation and Inference with Moment Methods for Dynamic Panels with Fixed Effects This chapter demonstrates the usefulness of the TSD estimator when applying to to dynamic panel datamodels. We find again that the TSD estimator displays a standard Gaussian limit, with a convergence rate of $\sqrt{nT}$ for all values of ρ, including unity, irrespective of how n or T approaches infinity. Particularly, the TSD estimator makes use of moment conditions that are strong for all values of ρ, and therefore can completely avoid the weak instrument problem for ρ in the vicinity of unity, and has virtually no finite sample bias. As in the time series case, the asymptotic variance of the TSD estimator is smaller than that of the TFD estimator of Han and Phillips (2009) when ρ > 0 and T > 3, and the corresponding t -ratio test is thus more capable of unveiling the true data generating process. Furthermore, the asymptotic theory can be applied directly to panel unit root test. Our simulation results reveal that the TSD-based unit root test is more powerful than the widely used IPS test (Im et al, 1997, 2003) when n is large and T is small.

In this paper, I analyze the effectiveness of state level Earned Income Tax Credit programs on improving of poverty levels. I conducted this analysis for the years 1991 through 2011 using a panel data model with fixed effects. The main independent variables of interest were the state and federal EITC rates, minimum wage, gross state product, population, and unemployment all by state. I determined increases to the state EITC rates provided only a slight decrease to both the overall white below-poverty population and the corresponding white childhood population under 18, while both the overall and the under-18 black population for this category realized moderate decreases in their poverty rates for the same time period. I also provide a comparison of the effectiveness of the state level EITCs and minimum wage at the state level over the same time period on these select demographic groups.

Coordenação de Aperfeiçoamento de Pessoal de Nível Superior / O objetivo principal desta dissertação é analisar os impactos da abertura comercial vista no Brasil no início da década de 1990 entre 1990 a 1996 na margem de lucro das indústrias de transformação passando por três padrões monetários diferentes (cruzeiro, cruzeiro real e real). A especificação e metodologia adotadas no trabalho são as de Goldar e Aggawal (2004), que fazem uma análise de dados em painel para efeitos fixos e randômicos para as indústrias de transformação indianas como um todo e, posteriormente, aplicando os mesmos testes separando os vinte e oito setores da indústria brasileira de transformação em setores de bens de capital e intermediários no primeiro grupo e bens de capital no segundo. Este trabalho ainda inclui esta metodologia aplicando, além das duas já citadas, o teste de mínimos quadrados ordinários (MQO) para permitir uma melhor análise com três testes diferentes antes e depois de separar os setores por meio de variáveis explicativas como barreiras à importação, concentração industrial, participação salarial, produtividade do trabalho, representatividade setorial e variação na taxa de crescimento da produção do setor entre os anos. Conclui-se que o aumento observado na margem de lucro foi impactado significativamente pelas variáveis expostas acima e estes resultados são importantes para que possamos auferir de que forma impactaram positivamente ou contribuíram negativamente na margem de lucro auferida pela indústria brasileira de transformação entre 1990 e 1996.

本研究探討選擇權交易量之大小對於波動度預測之效率性所造成之對偶效果(dual effect)，驗證〝正常的高交易量〞與〝異常的高交易量〞對於波動度預測能力是否有不同的影響。本研究採用panel data之資料型態，以LIFFE上市的個股買權為對象，資料長度為三年左右。主要欲探討之假說為： 1.一般而言，交易量大的選擇權，其波動度估計誤差較交易量小的選擇權來得小。 2.相對於平日水準而言，某日交易量異常高的選擇權將有較大的波動度估計誤差。 本研究所使用的波動度預測模型為隱含波動度(ISD)，採用的是最接近到期月份及最接近價平的合約。實證以組合迴歸、固定效果模型、隨機效果模型分別估計之，加以比較。結果發現固定效果模型為較佳之解釋模型，然而結果顯示交易量的對偶效果並不明確影響波動度預測誤差，故推測有某種影響公司間差異的因素，即公司間之異質性，比相對交易量更容易影響波動度預測之誤差。另外，透過組間與組內效果之分析，發現不論是長期還是短期，由於公司間的異質性存在，使得相對交易量對於波動度預測誤差均無明顯影響。 / The purpose of this research is to study the dual effect on the efficiency of volatility forecasting which is caused by the volume of option market, with the intent to test whether〝normal high volume〞and〝abcdrmal high volume〞cause different results on the ability of volatility forecasting. The data used is in the form of panel data. It is drawn from LIFFE, and has a length of about three years. The hypotheses to be examined in this study are：1. High-average-volume options have smaller volatility forecasting errors than low-average-volume options; 2. Options have larger volatility forecasting errors on abcdrmally-high-volume days than on normal-volume days. In this research, volatility is forecasted by implied standard deviation (ISD) which is implied in the at-the-money and the nearest expiry month options. Pooled regression、fixed effect model、and random effect model methods were applied. The results show that the fixed effect model made the best analysis amongst the three models. However, the result does not support the hypotheses made above, which means that volume does not have much influence on volatility forecasting error. It is inferred that there exists some other factors which could cause the difference between firms, namely heterogeneity, and these factors have much more powerful influence over volatility forecasting error than volume. Finally, it was found that no matter for long run or short run, because of the existence of heterogeneity, relative volume doesn’t have obvious influence on volatility forecasting errors when analyzing the difference between the between-individual effect and the within-individual effect.

This study examines the fundamental determinants of bilateral trade flows between South Africa and BRIC countries. This is done by exploring the magnitude of exports among these countries. The Gravity model approach is used as the preferred theoretical framework in explaining and evaluating successfully the bilateral trade flows between South Africa and BRIC countries The empirical part of this study uses panel data methodology covering the time period 2000-2012 and incorporates the five BRICS economies in the sample. The results of the regressions are subject to panel diagnostic test procedures. The study reveals that, on the one hand, there are positive and significant relationships between South African export flows with the BRICs and distance, language dummy, the BRICs’ GDP, the BRICs’ openness and population in South Africa. On the other hand, GDP in South Africa, real exchange rate and time dummy are found to be negatively related to export flows. / Economics / M. Com. (Economics)

Coordenação de Aperfeiçoamento de Pessoal de Nível Superior / O objetivo principal desta dissertação é analisar os impactos da abertura comercial vista no Brasil no início da década de 1990 entre 1990 a 1996 na margem de lucro das indústrias de transformação passando por três padrões monetários diferentes (cruzeiro, cruzeiro real e real). A especificação e metodologia adotadas no trabalho são as de Goldar e Aggawal (2004), que fazem uma análise de dados em painel para efeitos fixos e randômicos para as indústrias de transformação indianas como um todo e, posteriormente, aplicando os mesmos testes separando os vinte e oito setores da indústria brasileira de transformação em setores de bens de capital e intermediários no primeiro grupo e bens de capital no segundo. Este trabalho ainda inclui esta metodologia aplicando, além das duas já citadas, o teste de mínimos quadrados ordinários (MQO) para permitir uma melhor análise com três testes diferentes antes e depois de separar os setores por meio de variáveis explicativas como barreiras à importação, concentração industrial, participação salarial, produtividade do trabalho, representatividade setorial e variação na taxa de crescimento da produção do setor entre os anos. Conclui-se que o aumento observado na margem de lucro foi impactado significativamente pelas variáveis expostas acima e estes resultados são importantes para que possamos auferir de que forma impactaram positivamente ou contribuíram negativamente na margem de lucro auferida pela indústria brasileira de transformação entre 1990 e 1996.