The first specific problem is that there is no analysis or the analysis is not in-depth. When we write an article, sometimes the teacher's evaluation is not analyzed, and it feels that this thing is just a list, or it is rehearsed according to certain procedures and programs; Either the analysis is not in-depth, as if this thing only interprets these estimated results and does not dig some information behind it. The main reason is that I don't know what empirical analysis is, especially I can't understand and grasp empirical analysis from the function. The reason for this problem is that when we don't analyze it or analyze it in depth, it is actually related to the unclear goal of doing this work. If we know that the purpose or goal of empirical analysis is to acquire new knowledge, then relatively speaking, the degree of non-analysis or in-depth analysis may be weakened.
Second, there is no concern about hypothesis testing.
The second specific problem is that there is no concern about hypothesis testing. In other words, some empirical analysis, especially the articles written by undergraduates, feels like a running account. From the front to the back, they don't know the purpose of doing this, the conditions of empirical analysis from the function, the specific content of analysis and verification, and the test hypothesis, and then they don't know the function, the specific action and the object of empirical analysis.
3. Do not pay attention to identification and inference
The third specific problem is not paying attention to identification and inference. This is mainly about the bottom line, that is, we made a hypothesis, but there seems to be two skins between empirical analysis and hypothesis, which has little relationship. Theoretical analysis is theoretical analysis and empirical analysis is empirical analysis. One of the reasons for this problem is that the research design of empirical analysis is not enough, that is to say, the hypothesis is not verified and implemented in the research design, and the subsequent empirical analysis is not done step by step according to a specific idea and step of the research design. Secondly, we don't have a good grasp of the theories and documents related to the estimation results, because in many cases, identification and inference can be completed not only by some statistical or quantitative things, but also by the analysis of other theories and the mutual confirmation of relevant documents.
4. There is no diagnostic test, or there is no serious explanation of the test ideas and test results.
The fourth specific problem is that there is no diagnostic test, or the idea and result of the test are not carefully explained. The main reason for this problem is that we don't know or master the relevant knowledge and technology. We can see that there are many general tests. For example, we need to use panel data model and houseman test to test whether this thing is a random effect or a fixed effect. Without these knowledge and techniques, the paper will either be useless or can't explain the test ideas and results clearly.
Verb (abbreviation of verb) lacks robustness test.
At present, there are many robustness tests, such as the form of transfer function, index, sample area and time period, all of which belong to robustness tests. I don't have this part because I don't know, I don't understand, that is to say, what is the use of this thing, or how to do it is not clear. Secondly, it is a problem that the execution is not reached. Sometimes I often do some robustness testing, and I find that the results are in conflict with the results of benchmark regression. It's hard for me to say clearly, so I got stuck in this place, or I simply cut off robustness testing.