During this week, we mainly studied the literature retrieval and reading skills, data processing methods and the basic theory of regression models. Thanks to my undergraduate degree in finance and software engineering, I am familiar with this knowledge. I also help my friends within my ability. When installing the R language, I provide related links and study manuals for the students to use. As shown in Figure 1, Figure 2, and Figure 3.
Research work is very tired. Because my undergraduate degree in the computer research group for more than a year, but do not learn about the economic and financial research fashion. The initial plan was to experience economy research and make a final decision on my PhD area of research. But on the way to American University, the domestic Mentor only gave me one day to make this significant decision. I decided to accept it after I fully thought about it. The main purpose of coming to Harvard is to experience the culture here and prepare for future exchange study during the PhD and post-doctoral work.
This week we mainly studied data processing, modeling and testing of R and Tata. In general, I can complete the teacher's task well and ahead of time. And share my resources with my friends.
But I also encountered some difficulties. For example, I don't know how to deal with data (such as saliency) in economics. During my undergraduate computer research, data processing was completely another set of paradigms that could not be converted in time. Since I didn't understand it at the beginning, I chose the wrong variable based on the first attempt. However, under the guidance of professor, the experiment was re-run. I am currently working on paper writing.
In this week, we mainly completed the writing of the first paper and the work of Project 3. In the process of writing a thesis, professor gave us many useful suggestions, such as the structure of the paper, the logic of writing, and so on. Under the guidance of professor, I successfully completed the writing of the first paper. And I am honored to receive professor's only paper guidance in class. professor pointed out that my paper structure is relatively complete, but there are some small details that still need to be improved, such as some tables that are slightly thin and can be merged. I am very grateful for his guidance and made corresponding changes, the final version is shown in Figure 1.
Besides, I wrote a review on the second project concerning overview, strengths, weakness, and additional comments, providing new ideas to this research.
We also finished the project 3: CMTO. This empirical project analyzes data from pilot studies conducted in partnership with the King County Public Housing Authority (KCHA) and the Seattle Public Housing Authority (SHA). Families willing to accept the severs from CMTO are randomly divided into control group and experimental treatment group. The results show that being randomly assigned to receive the treatment increases the probability of moving to a high opportunity neighborhood by 18 percentage points while treatment on the treated increases 23%. There is evidence of treatment effect heterogeneity by Public Housing Authority, and lack of evidence of treatment effect heterogeneity by family income.
Next week is the last week of my research project. I intend to present to professor the idea that I hope to return to university as a PhD in the long run and even in the future. I know this is a difficult road, but I am still young, I have time and confidence to keep on going.
The fourth week was a tough but happy week. We had to complete the writing of the second paper in a week, which is not an easy job. Thanks to the applied econometrics courses I have studied before, I finished the task smoothly. Besides the panel VAR of the original paper, impulse response function analysis and variance decomposition, I added the ADF unit root test and cointegration test. I started to enter the experimental part from Tuesday. In order to draw a solid conclusion, I continued to test the relevant experimental content until Wednesday night. Although the ADF test results are very significant, the final cointegration test results have not been very satisfactory, and the data has a strong cointegration, which cannot be solved due to time limitation. It took another night to finish the paper and edit the layout two hours before the deadline.
In the final review, professor showed two other students, basically only two or three pages of content without typesetting. In fact, I had doubts before, I had been very difficult to complete in a few days and took days and nights to work on it this week, why other students can easily complete. Although the version I finally submitted is also in a hurry, the conclusions are very rough, there are many places to improve the writing and content, but I can at least achieve the length, basic content, and typesetting requirements. However, my classmates also have a lot of places worth my learning. In the presentation, the other group showed the results of the post-grouping experiments that they achieved. This kind of experiment is very rare and worth learning. Our group highlighted future research proposal and time planning.
After the last lesson, we made a simple summary and goodbye. Everyone sincerely wished each other a promising future. After that, professor and I had a talk in the conference room of the economics department. professor encouraged me that realizing the dream of having a Ph.D. degree in the United States is still hopeful, although there may be difficulties. I expressed my most sincere gratitude to professor, as well as the desire to be in touch with in the future and to do some work for professor. In this trip, I not only gained experience, research methods, friendship but also found my future development direction. The most important thing is to enjoy a happy time with the excellent professor.
I am appreciated that I have the opportunity to have the economic research experience at American University this summer. My professor proficient in teaching content, and thoroughly explained the courses (with depth). professor's course is associated with economic and financial disciplines, combined with computer technology, broadening the way of thinking (with broadness). Professor introduced novel developments and new achievements of this discipline, giving us as much information as possible, as well as the introduction of the trend and prospects (with density).
Empirical Research on Financing Constraints and Investment Behavior of Chinese Listed Companies: China's financial system is undergoing reform and innovation in the financial system during the transition period. However, the marginal capital-output rate has risen sharply in recent years, indicating that China's capital or financial efficiency is deteriorating. We use a VAR model to analyze the impacts fundamental factors of marginal productivity on Tobin Q with a panel data of listed companies of China from 1997-2008. The results show that the availability of internal funds is more important in explaining investment as well as the evidence of inefficient allocation of capital and that slower growth rates in China. This paper provides microscope insights for further deepening reform of China's financial system.
中国上市公司融资约束与投资行为的实证研究：转型期中国金融体系正在进行金融体制改革与创新。然而，近年来边际资本产出率急剧上升，表明中国的资本或金融效率正在恶化。我们使用VAR模型分析了托宾Q边际生产力的影响因素，以及1997 - 2008年中国上市公司的面板数据。结果表明，内部资金的可用性对于解释投资以及资本配置效率低下和中国经济增长放缓的证据更为重要。本文为进一步深化中国金融体系改革提供了显微镜见解。
Besides, the projects K given trained my programming, data analysis, and logical thinking skills. In project 1, we explored the upward mobility through data analysis and data visualization. We found the pooling average upward mobility of San Francisco (38405.54) is higher than the atlas's (34311.68) and children from low-income families do well generally have better outcomes for those from high-income families. In project 2, we checked whether the class size has an impact on students score by data plot with threshold and regression analysis. In project 3, we conducted an empirical project to analyze data from pilot studies conducted in partnership with the King County Public Housing Authority (KCHA) and the Seattle Public Housing Authority (SHA). Families willing to accept the severs from CMTO are randomly divided into the control group and experimental treatment group. The results showed that being randomly assigned to receive the treatment increases the probability of moving to a high opportunity neighborhood by 18 percentage points while treatment on the treated increases by 23%. There is evidence of treatment effect heterogeneity by the Public Housing Authority, and lack of evidence of treatment effect heterogeneity by family income. In project 4, we use variables from Google DataCommons to predict intergenerational mobility using machine learning methods. The measure of intergenerational mobility that we will focus on is the mean rank of a child whose parents were at the 25th percentile of the national income distribution in each county (kfr_pooled_p25). My goal is to construct the best predictions of this outcome using other variables, an important step in creating forecasts of upward mobility that could be used for future generations before data on their outcomes become available. We found Random Forest performed the best in this dataset. Random Forest has many strengths: on many current datasets, it has great advantages over other algorithms and performs well; it can handle very high dimensional (feature a lot) data, and no need to make feature selection (Feature subsets are randomly selected); After the training, it can give which features are more important; When creating a random forest, the generalization error without bias estimation and the model generalization ability is strong.; Training speed is fast, easy to make parallelization method; In the training process, can detect the interaction between the features; The implementation is relatively simple; For unbalanced data sets, it balances errors; If a large part of the features is lost, accuracy can still be maintained. However, Random forests have been proven to be over-specified on some noisy classifications or regression problems; For data with different values, attributes with more values will have a greater impact on random forests, so the attribute weights generated by random forests on such data are not credible. Therefore, we need to test it on other latest upgraded models (i.e. LightGBM or deep neural network), and we can also propose a novel to improve the performance.
In sum, thanks to professor again for cultivating an academic aptitude for independent thinking and an interest in economic research.
She was a highly engaged and active learner. In our program, I mainly introduced data processing, modeling and testing of R. Miss. S could always complete the tasks I assigned ahead of time, in a high quality. Also, she never hesitated to share her resources with her peers. Miss. S also impressed me deeply with her enthusiasm in academic discussions. For example, while processing economic data, she was confronted with an urgent problem. Despite of continuous efforts, she failed to solve the problem. Thus, she turned to me for guidance. Her open mind enabled her to solve most problems creatively.
Miss. S's final research was concerned to the financing constraints and investment behavior of Chinese listed companies. During the process, Miss. S excelled from a cohort of participators of our program. The structure of her paper was complete, and the logic of writing was very clear. In this project, she adopted a VAR model to make analysis, which was always comprehensive, including strengths, weakness, and additional comments as well as new ideas to the research. It was quite challenging but Miss. S worked it out with her academic intellect and independent thinking ability.
During the program, Miss. S communicated with me frequently. She was quite clear about her academic interest, and expressed her ideas about pursuing a master degree. Her enthusiasm and dedication along with various notable abilities undoubtedly qualify her potential in furthering her study and research. I am convinced that she is a good fit of your postgraduate program and she will improve even faster if cultivated in your vigorous program. I therefore would like to provide her with my sincere endorsement. For any possible questions you would like to ask, please feel free to contact me. I take pleasure in helping the great people!
Beijing Ph.D Village
Education Technology Co., Ltd
As the most energetic
provider of research educational services in China and North America, Ph.D
village offers education for helping stuedents to get into better colledge and
helping improve their quality of life. Our wide range of educational programs,
services and products includes research program, elite program, visiting
scholar program, as well as paper online education.
Ph.D village has become one
of the most recognized brand in research education area.
Since operation, the team
members of Ph.D village has experience of serving over 1000 students
上海之窗版权所有 未经允许 请勿复制或镜像 Copyright © 2012-2017 http://www.667la.com/, All rights reserved.