Working Papers

The article investigates the usage of Informer architecture for building automated trading strategies for high frequency Bitcoin data. Three strategies using Informer model with different loss functions: Root Mean Squared Error (RMSE), Generalized Mean Absolute Directional Loss (GMADL) and Quantile loss, are proposed and evaluated against the Buy and Hold benchmark and two benchmark strategies based on technical indicators. The evaluation is conducted using data of various frequencies: 5 minute, 15 minute, and 30 minute intervals, over the 6 different periods. Although the Informer-based model with Quantile loss did not outperform the benchmark, two other models achieved better results. The performance of the model using RMSE loss worsens when used with higher frequency data while the model that uses novel GMADL loss function is benefiting from higher frequency data and when trained on 5 minute interval it beat all the other strategies on most of the testing periods. The primary contribution of this study is the application and assessment of the RMSE, GMADL and Quantile loss functions with the Informer model to forecast future returns, subsequently using these forecasts to develop automated trading strategies. The research provides evidence that employing an Informer model trained with the GMADL loss function can result in superior trading outcomes compared to the buy-and-hold approach.

This study investigates the distributional impacts of carbon taxes, traditionally examined through simulation studies on the regressivity of hypothetical tax scenarios. However, the dy-namic influence of actually implemented carbon taxes on consumption/income poverty and inequality in a cross-country setting has been less scrutinised. This paper assesses the effect of carbon taxes introduced in the past three decades in 15 European countries on consumption shares of the lowest decile groups, poverty rates and inequality indices. The analysis shows that a $40/ton CO2 tax covering 30% of emissions leads to a consumption share increase of up to 4% for the bottom 20% and 40% of the population, a trend that persisted for five years post-implementation, particularly in nations that efficiently redistribute carbon tax revenues. This resulted in a modest reduction in consumption inequality over three years. In contrast, the impact of carbon taxes on income poverty and inequality is not statistically significant. These findings suggest that concerns about poverty and inequality due to carbon taxes can be miti-gated by implementing a moderate tax combined with a strategically efficient revenue redis-tribution mechanism.

This paper examines Eurobarometer survey data from 27,438 individuals across 28 EU Member States in 2019 to evaluate the awareness and impact of EU Energy Labels. Specifically, we analyze the role of socioeconomic characteristics such as age, gender, education, financial stability, and political engagement. Our results suggest that individual characteristics have a greater effect on the influence of labels on purchase decisions than on label awareness. However, significant heterogeneity across countries persists even after controlling for individual characteristics. Using our model, we conduct three exercises in which we assume a policymaker can either increase label awareness among all unaware individuals or target those with specific characteristics, and we demonstrate the resulting impact on the share of people whose purchases are influenced by the label. The findings reveal that even when label awareness is at its highest level, it does not necessarily translate into substantially higher influence on purchasing decisions in some countries. Additionally, at the country level, certain socioeconomic and political variables are positively correlated with label awareness.
 

In this paper, we analyze the determinants of individual’s willingness to pay higher prices and taxes and to reduce their standard of living to support environmental protection. Using data from the 2020 International Social Survey Programme (ISSP), Environment IV module from 26 countries on about 29,000 individuals, we investigate the influence of socio-demographic factors, consumer behavior, environmental beliefs, opinions, and attitudes. The findings reveal significant variations in willingness to bear financial burdens for environmental protection across different countries and socio-economic groups. Our analysis highlights the critical role of education, religion, political affiliation, and trust in institutions in shaping environmental attitudes and behaviors. Moreover, after controlling for individual characteristics, significant international disparities persist, with countries like India showing exceptionally high willingness across all measures, while many European countries, despite their progressive environmental policies, show lower willingness for higher taxes due to possibly already high tax burdens. These findings underscore the importance of tailoring policy communications to different socio-economic groups, emphasizing both the immediate and long-term benefits of environmental protection to enhance acceptance among various demographic segments.

We study substitution between fixed and mobile broadband services in South Africa using survey data on 134,000 individuals collected between 2009 and 2014. In our discrete-choice model, individuals choose fixed or mobile voice and data services in a framework that allows these services to be considered substitutes or complements. We find that there is substantial heterogeneity in the perception of these services as substitutes/complements. We use our model to simulate the uptake of fixed and mobile broadband across various demographic groups under different policy interventions, including: (i) a reduction in mobile data prices; (ii) an expansion in fixed-line coverage; (iii) a widespread distribution of computers; and (iv) broader internet access in schools and workplaces. Our results suggest that, when applied in isolation, these interventions do not significantly increase internet access among poorer households. In particular, the uptake of fixed broadband would remain limited, even if accessible to all households. This is because many households prefer mobile internet access, perceiving it as a substitute for fixed broadband.
 

In this paper, we investigate the effects of non-exclusive agreements between networks of mobile money agents on mobile network operator choices, using survey data from Tanzania conducted in 2017. By combining survey responses with geo-location data and information on agent proximity, we employ discrete choice models to analyze consumers’ decisions in subscribing to mobile network operators and their corresponding mobile money providers. Our findings highlight the significant influence of the distance to mobile money agents on consumers’ subscription choices. To explore the impact of interoperability (non-exclusivity) at the mobile money agent level, where consumers can use the nearest agent from any mobile money provider, we assess its effects on market shares of mobile network operators. Our results indicate that interoperability at the agent level has only a minor impact on market shares. Smaller operators experience marginal gains as their consumers can now utilize agents of larger providers, which are often closer in proximity. In conclusion, we find that interoperability at the agent level does not considerably alter the market structure in the context Tanzania during the period under consideration.
 

In this paper we analyse whether having a mobile phone impacts chances of getting employed. We use five waves of panel data from the National Income Dynamic Survey (NIDS), which was conducted in South Africa between years 2008 and 2017. In the estimation we include a vector of observable individual and household characteristics and account for unobserved heterogeneity amongst individuals. The estimation results suggest that mobile phone ownership has a positive impact on the change in employment status from unemployed to employed. On the other hand, ownership of a computer by a household and computer literacy do not increase the likelihood of getting employed. The average probability of becoming employed increases from 54.2% when no one among unemployed adults has a mobile phone to 57.4% when all of them have a mobile phone, which is an increase of 5.9%.

In this paper, we utilize survey data collected in 2017 from 12,735 individuals across nine Sub- Saharan African countries. We merge the survey data with geographic information related to the proximity of mobile network towers and banking facilities, based on the geo-locations of the respondents. Our estimation approach comprises a two-stage model. In the first stage, consumers make choices between adopting a feature phone or a smartphone. In the second stage, they make decisions regarding the use of mobile money services. Our findings reveal that network coverage significantly influences the adoption of mobile phones. Moreover, we observe that mobile money services are more favored by younger and relatively wealthier individuals for sending money, while older individuals and those with lower incomes tend to use mobile wallets for receiving money. Consequently, mobile money services facilitate younger migrant workers residing in areas with better infrastructure in providing support to their older relatives in less developed regions.

A large literature in behavioral science suggests that people’s emotional condition can have an impact on their choices. We consider how people’s emotions affect their stated preferences and willingness to pay for changes in environmental quality, focusing on the effects of incidental emotions. We use videos to induce emotional states and test the replicability of the results reported in Hanley et al. (2017). Additionally, we employ Face Reader software to verify whether the intended emotional states were successfully induced in our experimental treatments. We find that our treatments succeed in implementing the predicted emotional condition in terms of self-reported emotions, but had a variable effect on measured (estimated) emotional states. We replicate the key result from Hanley et al. (2017): induced emotional state has no significant effect on stated preference estimates or on willingness to pay for environmental quality changes. Moreover, we confirm that, irrespective of the treatment assignment or emotional state - be it self-reported or measured - we observe no significant effect of emotion on stated preferences. We conclude that stated preference estimates for environmental change are unaffected by changes in incidental emotions, and that preference estimates are robust to the emotional state of the responder.

This study investigates the impact of investor sentiment on stock returns and trading volume, challenging the efficient market hypothesis. Using CRSP data from May 1998 to March 2022, methods like Fama-MacBeth and quantile regression were applied to analyze sentiment indicators such as the VIX, AAII Investor Sentiment Survey, Consumer Confidence, and Baker-Wurgler Index. The findings reveal that investor sentiment significantly influences stock returns and trading volume, especially during uncertain times. Sentiment also affects financial metrics like SMB, HML, RMW, and CMA uniquely. This research provides new insights and practical implications for investors and analysts, emphasizing the importance of considering sentiment in investment strategies to better anticipate market movements and manage risks.

Research background: The number of research regarding employment effects of minimum wages is enormous. Another problem examined by prior studies is the impact of minimum wage increases on the wages. The evidence shows that minimum wage increases compress the wage distribution. The same literature brings conflicting evidence regarding minimum wage spill-over effects.
Purpose of the article: The study analyses the effects of a minimum wage increase on the wage distribution of low- and high-wage sectors and possible spill-overs. The analysed period 2014-2018 is characterized by relatively stable economic conditions, while the minimum wage increased by 25%. 
Methods: We follow case study method and as example Poland, the EU country with high share of minimum wage workers. We use individual data on wages and worker characteristics from the Structure of Earnings Survey in Poland for 2014–2018. We use reweighting and decompose counterfactual wage distribution.
Findings & value added: In low-wage sector, a wage increase in the left tail of the distribution is almost entirely due to the increase in the minimum wage level and spill-over effects are present throughout the distribution. In high-wage sector the role of the minimum wage growth is weaker and also the workers’ characteristics have substantial impact on wages; no spill-over effects of a minimum wage increase are observed. We demonstrate that the conflicting evidence on the effects of minimum wage changes on the wage distribution may occur because the effects differ across the low- and high-paid economic sectors. They depend on sector productivity and openness.

This study utilizes machine learning algorithms to analyze and organize knowledge in the field of algorithmic trading, based on filtering 136 million research papers to 14,342 articles ranging from 1956 to Q1 2020. We compare previously used practices such as keyword-based algorithms and embedding techniques with state-of-the-art dimension reduction and clustering for topic modeling method (BERTopic) to compare the popularity and evolution of different approaches and themes. We show new possibilities created by the last iteration of Large Language Models (LLM) like ChatGPT. The analysis reveals that the number of research articles on algorithmic trading is increasing faster than the overall number of papers. The stocks and main indices comprise more than half of all assets considered, but the growing trend in some classes is much stronger (e.g. cryptocurrencies). Machine learning models have become the most popular methods nowadays, but they are often flawed compared to seemingly simpler techniques. The study demonstrates the usefulness of Natural Language Processing in asking intricate questions about analyzed articles, like comparing the efficiency of different models. We demonstrate the efficiency of LLMs in refining datasets. Our research shows that by breaking tasks into smaller ones and adding reasoning steps, we can effectively address complex questions supported by case analyses.

In this study, we explore the dynamics of consumer choices in the Polish telecommunications market, focusing on preferences and valuations for home fixed, home mobile, and purely mobile internet connections. Key attributes such as speed, latency, data limits, and cost are examined. Central to our research is the investigation of how the integration of 5G technology might influence demand elasticity. Using a detailed discrete choice experiment, we apply a mixed logit model with random parameters to analyze stated choice data, enabling us to unravel the complexities of demand elasticity, especially in terms of own- and cross-price elasticities. This approach facilitates an assessment of the degree of substitutability between fixed and mobile internet services.
Our findings indicate a moderate substitution effect between fixed and mobile internet services. Results from a Small but Significant and Non-transitory Increase in Price (SSNIP) test suggest that these markets should continue to be regulated separately, mirroring the distinct regulation observed in fixed and mobile telephony. Furthermore, simulations provide insights into potential future market shifts with the advent of 5G services. This paper contributes significantly to the discourse on fixed-mobile internet substitution and offers vital insights for defining markets in antitrust discussions, competitive agreements, and potential mergers within the telecom sector.

This research presents a comprehensive evaluation of systematic index option-writing strategies, focusing on S&P500 index options. We compare the performance of hedging strategies using the Black-Scholes-Merton (BSM) model and the Variance-Gamma (VG) model, emphasizing varying moneyness levels and different sizing methods based on delta and the VIX Index. The study employs 1-minute data of S&P500 index options and index quotes spanning from 2018 to 2023. The analysis benchmarks hedged strategies against buy-and-hold and naked option-writing strategies, with a focus on risk-adjusted performance metrics including transaction costs. Portfolio delta approximations are derived using implied volatility for the BSM model and market-calibrated parameters for the VG model. Key findings reveal that system atic option-writing strategies can potentially yield superior returns compared to buy-and-hold benchmarks. The BSM model generally provided better hedging outcomes than the VG model, although the VG model showed profitability in certain naked strategies as a tool for position sizing. In terms of rehedging frequency, we found that intraday heding in 130-minute intervals provided both reliable protection against adverse market movements and a satisfactory returns profile.

Predicting the S&P 500 index's volatility is crucial for investors and financial analysts as it helps in assessing market risk and making informed investment decisions. Volatility represents the level of uncertainty or risk related to the size of changes in a security's value, making it an essential indicator for financial planning. This study explores four methods to improve the accuracy of volatility forecasts for the S&P 500: the established GARCH model, known for capturing historical volatility patterns; an LSTM network that utilizes past volatility and log returns; a hybrid LSTM-GARCH model that combines the strengths of both approaches; and an advanced version of the hybrid model that also factors in the VIX index to gauge market sentiment. This analysis is based on a daily dataset that includes data for S&P 500 and VIX index, covering the period from January 3, 2000, to December 21, 2023. Through rigorous testing and comparison, we found that machine learning approaches, particularly the hybrid LSTM models, significantly outperform the traditional GARCH model. The inclusion of the VIX index in the hybrid model further enhances its forecasting ability by incorporating real-time market sentiment. The results of this study offer valuable insights for achieving more accurate volatility predictions, enabling better risk management and strategic investment decisions in the volatile environment of the S&P 500.

The scope for the accurate calculation of the Loss Given Default (LGD) parameter is comprehensive in terms of financial data. In this research, we aim to explore methods for improving the approximation of realized LGD in conditions of limited access to the cash-flow data. We enhance the performance of the method which relies on the differences between exposure values (delta outstanding approach) by employing the machine learning (ML) techniques. The research utilizes the data from the mortgage portfolio of one of the European countries and assumes the close resemblance for similar economic contexts. It incorporates non-financial variables and macroeconomic data related to the housing market, improving the accuracy of loss severity approximation. The proposed methodology attempts to mitigate the country-specific (related to the local legal) or portfolio-specific factors in aim to show the general advantage of applying ML techniques, rather than case-specific relation. We developed an XGBoost model that does not rely on cash-flow data yet enhances the accuracy of realized LGD estimation compared to results obtained with the delta outstanding approach. A novel aspect of our work is the detailed exploration of the delta outstanding approach and the methodology for addressing conditions of limited access to cash-flow data through machine learning models.

In 2020-2021, several methodological changes were introduced in the Labour Force Survey (LFS), which caused disruptions in data analysis and inference: the Covid-19 pandemic forced a change in the data collection method, and from the beginning of 2021, planned changes related to the harmonisation of social surveys in the EU were introduced (changes in the subject and object coverage of the survey). The aim of this paper is to examine the impact of the methodological changes on the measurement of labour force participation in Poland. Based on the analysis of quarterly LFS data over the period Q1 2019. - Q4 2021, it is shown that the change in the recruitment and interviewing method to CATI and the change in the rotation scheme had a significant impact on survey selection, attrition, propensity to participate in person and thus also on the sample structure, and that the problems of survey selection are not fully compensated for in the process of generalising the results from the sample to the general population. By treating the change in survey method as a natural experiment, it has been shown that the method of recruitment affects the underlying results of the survey. Over the period Q3 2020 - Q3 2021, the changes introduced to the LFS together increased the estimates of the participation rate by around 0.6 percentage points, the employment rate by around 0.1 percentage points and the unemployment rate by around 0.9 percentage points relative to the pre-pandemic measures. If the effect of the inconsistent classification of some people as working in subsistence agriculture is also taken into account, the overestimation of the participation rate under the new methodology would be around 0.9 percentage points.

This study aimed to apply the algorithmic trading strategy on major foreign exchange pairs and compare the performances of machine learning-based strategies and traditional trend-following strategies with benchmark strategies. It differs from other studies in that it considered a wide variety of cases including different foreign exchange pairs, return methods, data frequency, and individual and integrated trading strategies. Ridge regression, KNN, RF, XGBoost, GBDT, ANN, LSTM, and GRU models were used for the machine learning-based strategy, while the MA cross strategy was employed for the trend-following strategy. Backtests were performed on 6 major pairs in the period from January 1, 2000, to June 30, 2023, and daily, and intraday data were used. The Sharpe ratio was considered as a metric used to refer to economic significance, and the independent t-test was used to determine statistical significance. The general findings of the study suggested that the currency market has become more efficient. The rise in efficiency is probably caused by the fact that more algorithms are being used in this market, and information spreads much faster. Instead of finding one trading strategy that works well on all major foreign exchange pairs, our study showed it’s possible to find an effective algorithmic trading strategy that generates a more effective trading signal in each specific case.

The study seeks to develop an effective strategy based on the novel framework of statistical arbitrage based on graph clustering algorithms. Amalgamation of quantitative and machine learning methods, including the Kelly criterion, and an ensemble of machine learning classifiers have been used to improve risk-adjusted returns and increase the immunity to transaction costs over existing approaches.  The study seeks to provide an integrated approach to optimal signal detection and risk management. As a part of this approach, innovative ways of optimizing take profit and stop loss functions for daily frequency trading strategies have been proposed and tested. All of the tested approaches outperformed appropriate benchmarks. The best combinations of the techniques and parameters demonstrated significantly better performance metrics than the relevant benchmarks. The results have been obtained under the assumption of realistic transaction costs, but are sensitive to the changes of some key parameters

This study explores how various work and family-related contexts moderated the link between work-from-home (WFH) and self-perceived changes to the career prospects among employees with children after over a year of the COVID-19 pandemic. We argue that the link between WFH and the perception of changes to one’s career prospects is likely to differ depending on gender, occupation, whether the employee has worked from home before the pandemic, how much time their children spent at home due to pandemic restrictions and the cohabiting status of the parent. We conducted fixed effects multinomial regression models using a unique multi-country dataset, including representative samples of parents with dependent children from Canada, Germany, Italy, Poland, Sweden, and the US. Employees with children who had prior experience with WFH before the pandemic were more likely to report improved career prospects than those who worked solely in the office. The positive effect of WFH for newcomers to the world of remote work was less unequivocal and varied based on occupation and gender. We also find that the presence of children at home and the cohabitation status substantially moderate the link between WFH and perceived changes to one's career prospects, with different implications based on the employee's gender. We fill the research gap by showing how fluid workers perceptions of career prospects depend on varying professional (prior experience with WFH and occupation) and personal (increased family demands) situations. This study also indicates the need for context-sensitive career management in organisations.