Using an Artificial Neural Networks Approach to Assess the Links Among Environmental Protection Expenditure, Energy use and Growth in Finland

This study uses an artificial neural network (ANN) to forecast Finland's economic growth from 1990–2020. Environmental protection expenditure (EPE), capital accumulation (CAPITAL), labor force (LABOR), renewable energy use (REC) and non-renewable energy use (NREC) that play an essential role in contributing to economic growth are selected as input variables. The variable to be estimated is per capita income (GDP). Data augmentation is made using the data between 1990 and 2020 by predicting the future with the ARIMA statistical model. As a result, the data obtained as only 30 data has been extended to 511 data. The generated data is set to be 80% training and validation data and 20% test data. The training data is used in the training process of the ANN model with five inputs and one output layer. The error values of the designed ANN model for the GDP value forecast are examined. To better analyze the performance of the ANN model, two studies, the best and the worst, are presented. The prediction functions found from the ANN structure for both studies are mathematically presented in matrix form through hyperbolic tangent functions consisting of the dependent variable by GDP and the independent variables by others. The answer to which of the other variables is effective in maximizing the GDP value was investigated with the linear programming maximization problem obtained from these equations. Thanks to these problems analyzed separately for the best and worst performance of ANN, it has been seen that the values EPE, REC and NREC have positive effects on maximizing GDP. Finally, the relationship between GDP and the other five variables is examined by considering the 102 data used for the test in the ANN. GDP has been shown to have a strong positive relationship between EPE, REC and NREC and a very weak negative relationship between CAPITAL and LABOR.