Pär Gunnar Victor StockhammarUniversitetslektor

För närvarande undervisar jag på kursen i tidsserieanalys och är examinator för kandidatuppsatserna.

Selected publications in international journals:

• Gustafsson, O., Villani, M. and Stockhammar, P. (2023), "Bayesian Optimization of Hyperparameters from Noisy Marginal likelihood Estimates", Journal of Applied Econometrics, 38, 577-595.
• Lindholm, U., Mossfeldt, M. and Stockhammar, P. (2020), "Forecasting Inflation in Sweden”, Economia Politica, 37, 39-68.
• Gustafsson, O. and Stockhammar, P. (2019), "Variance Stabilizing Filters", Communications in Statistics – Theory and Methods, 48, 6155-6168.
• Stockhammar, P. and Österholm, P. (2018), "Do Inflation Expectations Granger Cause Inflation?", Economia Politica, 35, 403-431.
• Stockhammar, P. and Österholm, P. (2017), "The Impact of US Uncertainty Shocks on Small Open Economies", Open Economies Review, 28, 347-368.
• Gustafsson, P., Stockhammar, P. and Österholm, P. (2016), "Macroeconomic Effects of a decline in Housing Prices in Sweden", Journal of Policy Modeling, 38, 242-255.
• Stockhammar, P. and Österholm, P. (2016), "Effects of US Policy Uncertainty on Swedish GDP Growth", Empirical Economics, 50, 443-462.
• Ul Hassan, M. and Stockhammar, P. (2016), "Fitting Probability Distributions to Economic Growth: a Maximum Likelihood Approach", Journal of Applied Statistics, 43, 1583-1603.
• Österholm, P. and Stockhammar, P. (2014), "The Euro Crisis and Swedish GDP growth – a Study of Spillovers", Applied Economics Letters, 21, 1105-1110.
• Stockhammar, P. and Öller, L-E. (2013), "Vad driver den ekonomiska tillväxten?", Ekonomiska samfundets tidskrift (Journal of the Economic Society of Finland), 2, 109-115.
• Stockhammar, P. and Öller, L-E. (2012), "A Simple Heteroscedasticity Removing Filter", Communications in Statistics – Theory and Methods, 41, 281-299.
• Stockhammar, P. and Öller, L-E. (2011), "On the Probability Distribution of Economic Growth", Journal of Applied Statistics, 38, 2023-2041.

Some potential PhD research ideas: 

1. AI-driven sentiment extraction for new leading economic indicators
This PhD project aims to use advanced AI methods—such as transformer-based neural networks, deep learning for natural language processing, and ensemble techniques—to analyze the vast streams of unstructured text from e.g. news, social media, protocols and financial reports. The objective is to extract sentiment and contextual signals that can be aggregated into novel leading economic indicators. These sentiment-derived indices would then be integrated into traditional nowcasting and forecasting frameworks, potentially improving the timeliness and accuracy of economic predictions by capturing early shifts in market perceptions and consumer confidence.

2. Hybrid ensemble models for real-time economic forecasting
This project would aim to develop an ensemble framework that integrates machine learning algorithms—such as random forests, convolutional neural networks (CNNs), and long short-term memory (LSTM) networks—to deliver a rapid and accurate current assessment of key economic variables. By harnessing heterogeneous data sources and capturing nonlinear patterns, the research intends to create adaptive forecasting systems that respond robustly to structural shifts in economic time series. This would potentially also improve the nowcasting and forecasting performance.

3. Assessing the impact of stationarity transformation choices on model estimation and forecasting
This study will systematically investigate how different statistically valid stationarity transformations—such as differencing, trend filters and seasonal adjustments filters affect model parameters, inference, and forecast accuracy in time series analysis. By combining theoretical exploration with simulation studies and empirical applications on e.g. economic and financial data, the project aims to provide a deeper understanding of not only the effects but also give guidance on how to select transformations in different situations.