Cluster Forecasting of Corruption Using Nonlinear Autoregressive with Exogenous Variables (NARX) – An Artificial Neural Network Analysis

Abstract

Any effort to combat corruption can benefit from an examination of past and projected worldwide trends. In this paper, we forecast the level of corruption in countries by integrating an artificial neural network modeling and time series analysis. The data were obtained from 113 countries from 2007 to 2017. The study is carried out at two levels: (a) global level where all countries are considered as a monolithic group; and (b) cluster level, where countries are placed into groups based on their development-related attributes. For each cluster, we use the findings from our previous study on the cluster analysis of global corruption using machine learning methods that identified the four most influential corruption factors, and we use those as independent variables. Then, using the identified influential factors, we forecast the level of corruption in each cluster using a nonlinear autoregressive recurrent neural network with exogenous inputs (NARX), an artificial neural network technique. The NARX models were developed for each cluster, with the objective function in terms of the Corruption Perceptions Index (CPI). For each model, the optimal neural network is determined by finetuning the hyperparameters. The analysis was repeated for all countries as a single group. The accuracy of the models is assessed by comparing the mean square errors (MSE) of the time series models. The results suggest that the NARX artificial neural network technique yields reliable future values of CPI globally or for each cluster of countries. This can assist policymakers and organizations in assessing the expected efficacies of their current or future corruption control policies from a global perspective as well as for groups of countries.

Cluster Forecasting of Corruption Using Nonlinear Autoregressive Models with Exogenous Variables (NARX)—An Artificial Neural Network Analysis

“…Multiple linear regression models or other similar methods are insufficient to model such complex systems, which exhibit non-linear relationships among crucial attributes and outcomes. Therefore, a method that can handle time series in complex systems is required to model and predict corruption [7]. Artificial neural networks (ANNs), machine learning algorithms, using data that have been processed in prior research using other machine learning techniques, are applied in this study due to their potential for solving problems of this nature [8][9][10][11].…”

“…There is rather limited data that can be used in studies of this nature. The data were from the following databases [7]: the World Bank Group (WBG) [28], the United Nations Department of Economic and Social Affairs (UNDESA) [29], the United Nations Development Program (UNDP) [30], the World Economic Forum (WEF) [31], and Transparency International (TI) [1] (see Table 1). All data are numerical.…”

“…More discussion on the reasoning behind choosing these attributes can be found in our recent research study on the cluster analysis of corruption level in continents using principal component analysis and machine learning techniques [7]. In this paper, a principal component analysis (PCA), cluster analysis, and a random forest technique are used to determine CPI values for countries.…”

“…The first is the global level (all countries considered together as a single group). Then, to ensure model flexibility by avoiding making the same predictions for countries that are very dissimilar in terms of development-related attributes, cluster-level analysis was carried out using techniques established in the literature [7]. The four most influential factors of corruption (measured in terms of Corruption Perceptions Index (CPI)) for each cluster identified in the previous study are the independent variables in the model in this paper.…”