Abstract:  It is generally believed that university presidents play a key role in the development of universities, and the cognitive and emotional characteristics of university presidents will influence the decision-making and performance of universities. This paper makes a investigation on the relationship between the tenure of university presidents and university performance in the past 20 years in China. The theoretical significance of this study is to test the relationship between the characteristics of the tenure of university presidents and university performance in of China, enriching the empirical study of Upper Echelons Theory on university-related issues. The practical significance of this study is that extending the tenure of university presidents is conducive to improve university performance. Based on the problems raised in this paper, it is hoped that the research results of this paper can promote the relevant research in the academic circles and provide a reference for the national education management department to formulate the term system of principals.In this paper, panel data are used for regression analysis. Inthe past researches, the cross-section research and the time series research method are mostly used. Compared with traditional cross section data or time series data, panel data has many advantages, which can provide a large amount of data for research, increase the degree of freedom of data, control the heterogeneity of different individuals, avoid multiple collinear problems more easily, and is more suitable for dynamic adjustment process. Therefore, the validity of econometric model estimation is improved. In this paper, the first-class universities determined by the Ministry of Education in 2017 are selected as research samples. Because the ranking of the Minzu University of China and Xinjiang University fluctuates greatly, it is treated as abnormal values. National University of Defense Technology is not included in the research scope because it belongs to military system. The total number of observations in this study is 39. Taking into account the ranking changes in each year from 1999 to 2018, after excluding missing data and abnormal values, the final number of observations is 663. The characteristics of university presidents in this paper include gender, age, overseas education background, overseas work experience, tenure, academician, internal promotion and alumni. The features of past and current university presidents are collected from the websites of universities and colleges, and the missing data are partly collected from the websites of Baidu Encyclopedia and Ministry of Education. The change of university performance manifested as change in ranking, which refers to the fluctuation when the university president takes office and leaves the university, which is derived from Wu Shulian comprehensive university ranking from 1999 to 2018. Because the performance of the university has lag effect compared with the president tenure, the decisions of university presidents generally can not produce significant effect in the current year. Therefore, in order to reflect causality, we choose the following year as the research object, and the performance of the following year will be published in the form of ranking at the beginning of the following year. Therefore, the independent variable will be measured by the t year of the university president in office, and the dependent variable will be measured by the university ranking published in t+2 year. The statistical results show that, among the presidents of first-class universities, those who have studied abroad accounted for 42.4%, those who have overseas work experience accounted for 33.2%, academicians accounted for 32.4%, and internal promotions accounted for 57%, the proportion of alumni accounts for 54.3%, the average tenure is 4.293 years. The following conclusions can be drawn from the correlation coefficient matrix: There is a significant positive correlation between the tenure of university presidents and university performance. The overseas work experience has passed the coefficient test under the 0.01 confidence level. Academician has passed the coefficient test under the 0.05 confidence level. Merge has passed the coefficient test under the 0.001 confidence level. It is shown that these three control variables can explain the university performance better, and adding these variables can help to analyze the relationship between the background characteristics of university presidents and university performance more accurately. But gender, age, overseas background, internal promotion, and alumni are less explanatory. The relationship between the tenure of the university president and university performance has passed the significant test under 0.05 confidence level, which indicates that the longer the tenure of the university president, the better the performance of university. The squared coefficient of thethe tenure of the university president is negative but not significant, and does not pass the significance test. We believe that the reason why the square of tenure does not pass the significance test is that, the term of office of university presidents in our country is generally short and has not reached the value for the highest highest performance level.First, the longer the tenure of the university president, the better the performance of university. Second, the square of tenure fails to pass the significance test, mainly because compared with foreign countries, the tenure of of university presidents in our country is generally shorter and has not reached the value for the highest highest performance level. This paper holds that, based on the conclusion of the research, the term of the university president should be extended properly for the long term development of the first class university in China.

Key words: university president, tenure, university performance, upper echelon theory