Search

Can artificial intelligence (AI) think and act like a cash manager? In this paper we explore how generative AI agents can help manage liquidity, prioritize payments and optimize efficiency in real-time gross settlement systems.

We study the cross-section of equilibrium returns on safe assets using a tractable asset pricing model with a micro-founded demand for liquidity and multiple safe assets with heterogeneous transaction costs. We test the main predictions of our theory using a novel measure of relative (in)convenience yields in the US Treasury market.

In 2024, the Canadian Overnight Repo Rate Average (CORRA) rose 7 basis points above the Bank of Canada’s target overnight rate as settlement balances declined and hedge fund borrowing increased by $30 billion, straining dealers’ balance sheets. Exercising market power, dealers raised rates, and as client activity grew, these higher rates increasingly influenced CORRAs deviation from target.

We study how participants in the Lynx payment system use the net send limit (NSL) tool to control their intraday payment outflow levels. Our results show that participants typically adopt a “set it and forget it” approach to scheduling NSLs and sometimes have distinct intraday NSL adjustment behaviours.

We examine the Canadian money market fund (MMF) sector and find that it has grown rapidly, holding a large share of treasury bills and commercial paper. Unlike in some other jurisdictions where investor outflows likely amplified stresses, Canadian MMFs experienced inflows during the March 2020 market turmoil.

We identify two channels that affect the relationship between inflation and unemployment. First, inflation lowers wages because unemployed suffer more from inflation than employed, generating a positive relationship. Second, inflation increases firms’ financing costs, generating a negative relationship. Improvements in firm financing conditions can induce the relationship to switch signs.

This paper studies how financial shocks shape the distribution of output growth by introducing a quantile-augmented vector autoregression (QAVAR), which integrates quantile regressions into a structural VAR framework. The QAVAR preserves standard shock identification while delivering flexible, nonparametric forecasts of conditional moments and tail risk measures for gross domestic product.

We study consumer cash inventory behavior by developing a dynamic model of forward-looking consumers and estimating structural parameters of the model using detailed consumer survey data. Consumers facing holding and withdrawal costs solve a discrete-time continuous-control dynamic programming problem to optimally use cash at the point of sale.

We present core findings from the 2024 Methods-of-Payment Survey, highlighting results from both the survey questionnaire and three-day shopping diary. Although cash holdings have increased in nominal terms, we find that cash usage remains unchanged since 2020. Mobile and other alternative payment methods continue to grow in importance.

We outline possible system architectures for an online retail central bank digital currency (CBDC) and identify one well-suited for basic payments. We analyze OpenCBDC 2PC, a representative system, finding such designs to be feasible, although some areas require further investigation. Our framing highlights other promising architectures for a retail CBDC.