computational economics
Nowcasting Macroeconomic Variables Using Payments Data
|
The recent COVID-19 crisis underscored the importance of having more frequent access to estimates reflecting the current state of the economy, as these estimates aid policymakers in directing monetary and other types of economic policy. This paper aims to determine whether payments data can be used to forecast ('nowcast') macroeconomic indicators in the near future, in order to avoid the lag that official gross domestic product (GDP) and consumer price index (CPI) figures have. Payments Canada has near-real-time access to the payment system variables, which cover a wide range of spending activities and allow for timely macroeconomic forecasting. We use machine learning (ML) techniques to build the nowcasting model. The paper presents preliminary findings on the use of payment data to improve GDP and CPI forecasts. We find that selected macroeconomic indexes and payments system data are co-integrated, and ML modeling demonstrates significant predictive ability for real GDP and CPI.
Keywords: nowcasting, payments data, machine learning, macroeconomic indicators, GDP, CPI.
Insights from the LVTS overnight loan market: implications for Lynx collateral
|
This study produces a series of stylized empirical facts about the overnight loan market activity in the Canadian high-value payment system. The in-depth exploration of overnight loans for the period of 2004–2020 was conducted using the Furfine algorithm. We find that almost all overnight loan activity between participants takes place via Tranche 2. The paper also provides a heterogeneity analysis based on the bank size and pair type. We suggest the relationship between a bank's size and its market activity and find that large banks tend to lend more frequently, but the average value of a loan is usually smaller compared to smaller banks. Our findings shed light on the downward trend in overnight market activity during financially unstable times, specifically the recession of 2008/2009 and the latest COVID-19 crisis. We also discover a time-of-day pattern in the market activity and conclude that the overnight market activity spikes prior to the end-of-day settlement. Further, we contribute to the better identification and measurement of payments data and its importance for monitoring systemically important payment systems.
Keywords: overnight loans, LVTS, Lynx, interbank lending, liquidity, global financial crisis, COVID-19
ARE CRYPTOCURRENCIES CRYPTIC OR A SOURCE FOR ARBITRAGE?
|
This paper aims to identify forex triangular arbitrage trading opportunities using a branch of evolutionary algorithms known as genetic algorithms (GAs) to derive insights into the volatility cryptocurrencies and stablecoins with the largest market cap. The triangular trade will be carried out as a 3 or more-tuple arbitrage play consisting of two or more cryptocurrencies and a fiat currency (the USD) that is used to enter and exit the trades. Our results show persistent tradable arbitrage for digital currencies of $2 for stablecoins and $5 for cryptocurrencies and $25 for stablecoin mix strategy. Triangular arbitrage trading yielded no arbitrage profits with the fiat currency strategy, as expected.
Keywords: cryptocurrency, arbitrage, genetic algorithms (GAs), evolutionary algorithms (EAs), trading strategies, optimization
PAYMENT STREAM VOLUME MIGRATION: TARGET STATE VOLUME PROJECTIONS
|
Forthcoming
In order to account for potential risks and implement relevant policies, this paper aims to estimate the directions of migrations between current and new payment systems under management by Payments Canada as well as each system’s market share for the next 5 years. Seasonal Auto-Regressive Integrated Moving Average (SARIMAX) modeling is used to predict the total volume of transactions for each payment stream. The Vector Autoregressive (VAR) model and the Ordinary Least Squares (OLS) models are selected to find the relationship between the calculated net migration volume and the total volume for different streams.
With economic and work-life adjustments resulting from the COVID-19 global pandemic, the analysis aims to account for potential structural changes to the economy that have resulted from the pandemic and affected payment ecosystem.
In order to account for potential risks and implement relevant policies, this paper aims to estimate the directions of migrations between current and new payment systems under management by Payments Canada as well as each system’s market share for the next 5 years. Seasonal Auto-Regressive Integrated Moving Average (SARIMAX) modeling is used to predict the total volume of transactions for each payment stream. The Vector Autoregressive (VAR) model and the Ordinary Least Squares (OLS) models are selected to find the relationship between the calculated net migration volume and the total volume for different streams.
With economic and work-life adjustments resulting from the COVID-19 global pandemic, the analysis aims to account for potential structural changes to the economy that have resulted from the pandemic and affected payment ecosystem.
Keywords: