The Impact of Government Performance Evaluation on Environmental Performance in Organizations

Abstract

Using GPE performance gaps (the difference between actual and standard levels of performance), this article explores the relationship between the GPE outcomes of public organizations and their environmental performance (EP). We utilize Korean GPE data related to qualitative/quantitative and financial/non-financial performance indicators that the government uses to evaluate public organizations. The GPE does not cover every important environmental activity in which public organizations may engage; the EP measure can incorporate results from all environmental activities, even those excluded from the GPE. Our investigation revealed that the EP of public organizations varied according to their GPE performance gaps. Low GPE performers favored GPE-included environmental activities over GPE-excluded environmental activities, but high GPE performers engaged in environmental activities were excluded from the GPE. A negativity bias occurs when public organizations adjust their environmental activities in response to GPE performance gaps. For environmental activities excluded from the GPE, low GPE performers avoided them more than high GPE performers engaged in them. These findings indicate the importance of a well-designed organizational performance evaluation for balanced engagement in environmental initiatives. This article is of theoretical interest to academics and has practical value for practitioners.

Keywords:

• Environmental performance (EP)
• government performance evaluation (GPE)
• negativity bias
• performance gaps

Notes

1 For example, all environmental indicators on raw material procurement and transportation, power generation, and distribution should be included in the GPE of a public energy provider.

2 In this article, low GPE performers are public organizations that receive low performance grades (D or E out of S, A, B, C, D, and E) on the Korean GPE, not those that receive low performance scores on the EPE. Similarly, high GPE performers are public organizations that receive high performance grades (S or A out of S, A, B, C, D, and E) on the Korean GPE, not those that receive high performance scores on the EPE. See details of Korean GPE grades in the section entitled, “The GPE and EPIs in Korean SOEs: The Context.”

3 The EPE can be defined as a “process to facilitate management decisions regarding an organization's environmental performance by selecting indicators, collecting and analyzing data, assessing information against environmental performance criteria, reporting and communicating, and periodically reviewing and improving this process” (ISO 14031, 2013). EPIs, or indicators that provide information about an organization's environmental performance, organize vast quantities of environmental data about the organization comprehensively and concisely (Jasch, 2000).

4 Behavioral theory examines performance gaps that may exist between an organization's satisfactory aspirations and reality from a behavioral perspective, demonstrating how the organization responds to these gaps through allocation of its resources (Hong, 2019; Hong et al., 2020; Yu et al., 2021). This theory presupposes that members of an organization have limited knowledge and rationality. All organizations look for acceptable alternatives, not necessarily the best ones. One viable option is to fulfill the organization's aspirations, whether historical or social in nature. Historical aspiration is motivated by the desire to outperform previous accomplishments. By contrast, social aspiration is the pursuit of superior performance to that of competing organizations. Which of these two goals the organization prioritizes is determined by their relative importance (Cyert & March, 1963; Greve, 1998). Strategic reference point theory (Fiegenbaum et al., 1996) is based on prospect theory (Kahneman & Tversky, 2013). It claims that managers of organizations choose aspirations to make their employees consider how they want to perform at work (Shinkle, 2012). Complementary interventions such as reward systems (Fiegenbaum et al., 1996) and punishment systems are implemented in accordance with their risk attitudes (Yu et al., 2021). For instance, when performance falls short of the desired level, risk-taking managers (Fiegenbaum et al., 1996) view organizational issues as opportunities for organizational change (Shinkle, 2012).

5 We exclude negativity dominance from our analysis. Negativity dominance occurs when combinations of negative and positive entities yield more negative evaluations than the algebraic sum of individual subjective valences would predict (Rozin & Royzman, 2001). However, low performance and high performance cannot occur simultaneously in an organization. Negative differentiation is also excluded from our analysis. It asserts that negative stimuli are considered as more recalled and memorable than positive stimuli. However, analysis of employees' memories or psychology goes beyond the scope of this article. Therefore, these two constructs are excluded from the analysis.

6 The EPE accounts for just one point out of a hundred points in the GPE. As a result, our model has a very low probability of collapsing or encountering a problem of dual causality.

7 To begin, the Hausman test revealed a correlation between the independent variable and the error term [Chi²(8)=1.78, p-value = 0.9870]. According to Greene (2011), the Hausman test compares the covariance matrix of the regressors in the fixed effects model with those in the random effects model, under the null hypothesis that there is no systematic difference between the two covariance matrices. The fixed effects model is superior to the random effects model if the null hypothesis is rejected. When panel data exhibit heteroscedasticity and autocorrelation, an error in the Hausman test result may occur. As a result of running the additional Robust Hausman Test to compensate for this problem, the null hypothesis that there is no systematic difference in the covariance matrix was not rejected, indicating that the random effects model was suitable [Chi²(8)=1.08, p-value = 0.9977]. Second, using the fixed effects method, variables that do not change over time are dropped, so there is a limit to the extent to which the effect may be estimated (Hill et al., 2020). For example, race, gender, and marital status, which are critical variables in social sciences research for explaining social phenomena, do not change over time, making it difficult to estimate the effects using the fixed effects method. Among the article's control variables, industry has a significant impact on the environmental performance of public agencies. This is because the amount of environmental pollution and the content of the regulatory policies in place vary by industry. When analyzing the industry of public agencies (a variable that does not change over time) using the fixed effects method, this variable is dropped, imposing a constraint on the estimation. Third, in time-series cross-sectional data, independent variables that change very gradually over time (e.g., government politics and political connection in this article), especially in comparison to changes in the dependent variable, are frequently referred to as slow-moving or sluggish. If the correlation between the sluggish covariate and the unit fixed effects is sufficiently high, this can significantly destabilize estimates of the independent variable's effect, resulting in extremely unreliable inferences. Random effects models are not subject to either of these limitations (Clark & Linzer, 2015, p. 9).

8 The Alio system, the main source of these data, provides information only for the last five years, during which time 37 Korean SOEs were in existence.

9 In order to assess the validity of our model, we also examined representative tests for heteroscedasticity and autocorrelation. Results of the White Test did not indicate heteroscedasticity (prob > chi²=0.2940). In addition, the Wooldridge Serial Correlation Test did not reveal an autocorrelation problem (prob>F = 0.2430).

10 Performance and integrity often tend to be managed separately because they are regarded as mutually exclusive. However, Ko (2015) showed that including an integrity indicator in the GPE induces a positive correlation between performance and integrity.

Additional information

Notes on contributors

Seungwon Yu

Seungwon Yu, PhD, is an associate professor in the Department of Public Administration, Korea National Police University, Korea. His research focuses broadly on political economy, public management, policy analysis and evaluation, and state-owned enterprises.

Ga-Hui Shin

Ga-Hui Shin, PhD, is a research associate at Seoul Institute, Korea. Her research focuses broadly on public management, policy analysis and evaluation, and local government.

Suhee Kim

Suhee Kim, PhD, is an assistant professor in the Department of Urban Administration, University of Seoul, Seoul, Korea. Her research focuses broadly on public management, local government, Korean politics, policy analysis and evaluation, and state-owned enterprises.