Forecasting egg price inflation in Germany with machine learning: A comparative study with ARIMAX and LSTM

Abstract The recent surge in egg prices in Germany and the US raises concerns, as eggs are a basic and sustainable food item. Previous studies on egg prices explored the influence of supply and demand factors. In addition, recent events such as the COVID-19 pandemic, the war in Ukraine, and avian influenza, as well as policy changes for animal welfare protection in Germany such as the ban on chick culling and cages, may have contributed to the price volatility of the egg market. Amidst this large number of potential features, this paper firstly sets out to identify the most important predictors of egg prices in Germany. In the second step, we aim to predict egg prices using the most relevant features. With the Least Absolute Shrinkage and Selection Operator, we identify mainly supply related factors, such as egg imports, the battery cage ban, the 2017 fipronil scandal, the 2022 ban on male chick culling, energy prices, the number of COVID-19 cases, the number of bird flu outbreaks and seasonal changes in demand during Christmas, as the most important determinants for changing egg prices. From the perspective of the ‘No Free Lunch Theorem’, regarding predicting egg prices, we compare the deep-learning nonlinear recurrent neural network represented by the Long Short-Term Memory (LSTM) to the linear Autoregressive Integrated Moving Average with Exogenous Variables (ARIMAX) model and find that the ARIMAX predictions consistently underperform compared to LSTM. The findings underline the crucial role of policymakers in stabilizing egg prices by effectively preventing and managing irregularities in the value chain.