高等教育效率评估：实证研究综述

需积分: 5 120 浏览量更新于2024-07-09 收藏 797KB PDF 举报

"这篇研究论文是对高等教育效率评估的综合回顾，涵盖了1997年至2018年的76项实证研究。作者Vanesa D’Elia和Gustavo Ferro对这些研究进行了分类，主要依据所采用的方法、输出、输入、质量和上下文变量的定义。大部分研究（72%）运用了非参数方法，而最新的研究倾向于使用面板数据。完成的学位作为最常见的输出变量被广泛使用，但仅有9篇论文考虑了质量变量。此外，虽然参数方法的论文中只有少数考虑了观察到的异质性，但有40%以上的参数模型纳入了环境变量以应对这种异质性。该综述对关注高等教育效率的研究人员、政策制定者及教育领域相关人员具有重要参考价值。" 在高等教育效率的测量中，各种方法和技术被广泛应用。非参数方法，如数据包络分析（Data Envelopment Analysis, DEA）和随机前沿分析（Stochastic Frontier Analysis, SFA），是评估效率的常用工具。DEA是一种非参数技术，能够处理多个输入和输出变量，不依赖于特定的函数形式，但可能忽视了效率变化的内在随机性。相反，SFA考虑了生产过程中的随机误差，允许分析效率损失的原因，如技术退化或不可控的外部因素。在这篇综述中，完成的学位被多数研究用作衡量高等教育机构效率的主要输出指标。然而，仅包含学位数量的评估可能不足以全面反映教育质量，因为质量是高等教育的重要组成部分。因此，只有9篇论文尝试量化并纳入质量变量，这表明在效率研究中对高等教育质量的衡量仍存在不足。另一方面，参数方法，如生产函数分析，通常用于考虑投入和产出之间的定量关系，但在处理异质性时可能面临挑战。为了克服这一问题，超过40%的参数研究选择包括环境变量，这有助于解释不同机构之间效率差异的观察到的异质性，例如地理位置、财政资源、学生背景等因素。这篇论文的重要性在于它提供了对过去二十年间高等教育效率研究的全面概述，这对于理解当前的研究趋势、方法论选择以及未来研究的潜在方向至关重要。对于政策制定者来说，这些信息有助于制定更有效的资源分配策略，以提高教育系统的整体效率。同时，对于研究人员，它可以指导他们设计更为严谨且全面的效率评估模型，考虑更多的变量，如质量、异质性和动态变化。

intensive services to keep rising year after year, in comparison with goods’ sectors whose growing

automation declines unit costs. The most recent source of change in production activity in the

universities is the increased use of information technology and e-learning. In teaching, these may

allow class sizes to increase, but an increasing ratio of students to staff may have an adverse effect on

technical efficiency (Johnes, 2007).

As Wolff et al. (2014) state, the productivity cannot be enhanced -while maintaining quality-,

by putting students on an assembly line or substituting robots for teachers, or making class sizes

larger and larger. It is not clear if technology-based approaches -such as e-learning- will yield

educational results that match those of current educational methods.

This study seeks to describe and summarize the most relevant empirical literature that

assesses efficiency in higher education. This review attempts to be context-setting and a useful

background for researchers, scholars who are outlining an empirical study and for policy makers

seeking informed criteria for decisions. In doing so, we analyze 76 specific studies ranging from 1997

to 2018 which assess higher education efficiency and classify them according to the methodologies

applied and to the definitions used to describe the outputs, inputs, quality and the context variables.

One previous study on this issue is the work of De Witte and López-Torres (2017) who provide a

survey of the literature on efficiency in education in general. In this paper we focus on higher

education efficiency.

The rest of the paper is organized as follows. Section 2 summarizes the different

methodological options applied in the studies reviewed, Section 3 analyzes the production/cost

drivers used to empirically compute efficiency. Finally, Section 4 summarizes the major contributions

of the analyzed literature, evaluating the current stage of the field.

2. Frontier Methods

There are two families of techniques for measuring efficiency: parametric or regression based, and

non-parametric or mathematical programming estimators. The mathematical programming methods

are generally deterministic (not distinguishing between pure randomness and efficiency) and non-

parametric (not assuming a functional form between the variables). They model the productive

process, but they do not estimate a function nor its parameters, in the sense of economic theory. The

most common non-parametric technique is Data Envelopment Analysis (DEA). This method

characterizes the set of efficient producers (those on the frontier), and then derives estimates of

efficiency for inefficient observations based on how far they deviate from the most efficient ones.

They are Total Factor Productivity measures which require finding some way to fairly assign weights,

or importance, to the various inputs and outputs included. The DEA method seeks to determine

which universities form an envelope surface with respect to the sample data. The units on the

frontier are considered efficient, while those below the envelope are considered inefficient.

The measurement of inefficiency is given by the distance between the individual university

and the frontier. In DEA there is no need to draw assumptions about efficiency a priori or even to

objective functions of the units under analysis (Salerno, 2003). Being a very sensitive method to

outliers, DEA gives warnings to detect unusual data in the sample, sometimes neglected in the

econometric work, especially in large samples.

For the introduction of environmental variables, there are several methods derived from the

DEA technical efficiency ranks such as two stages DEA and bootstrapping methods. Moreover, the

Malmquist index allows to expand the findings obtained from DEA, and to reveal changes in efficiency

scores and technical change over time (De Witte and López-Torres, 2017).

Electronic copy available at: https://ssrn.com/abstract=3532020

剩余20页未读，继续阅读

weixin_38716563

粉丝: 5

高等教育效率评估：实证研究综述

美国利率对汇率影响的实证研究：1998----2005

基于智能算法优化的LSSVM模型：GWO-LSSVM与AO-LSSVM及其DLHGWO-LSSVM优化研究,基于智能算法优化的LSSVM模型：GWO-LSSVM与AO-LSSVM的实证研究及DLHGW

企业风险披露：回顾-研究论文

大规模实证研究揭示：Wi-Fi连接耗时问题及其影响因素

台湾和泰国谷歌街景隐私：实证分析-研究论文

近期科技变革的技能内容：实证探索-研究论文

中国电子信息制造业产业聚集现状的实证研究：2004-2013年.pdf

包容性发展视角下中国省区全要素生产率的实证研究：1997-2012年

电子商务与供应链管理：概述和框架-研究论文

计量经济学遇上情绪：方法论和应用概述-研究论文

最新资源