Generative AI Can Produce Real-Time Deposit Pricing Strategies
By Olly Downs, Curinos
Simple Subscribe
Subscribe Now!
Executive Summary
- Deposit pricing tools have come a long way, but there’s a disconnect between what they produce and action that hits the market. Generative AI, handled properly, can accelerate implementation measurably.
- Pricing recommendations devised via GenAI can reach decision makers in hours, instead of weeks.
- Timely action can garner deposits when they’re needed at favorable costs, leaving competitors in the dust.
In just the past few years, deposit pricing has moved from competitive surveys and sheer intuition to optimization based on a breadth of industry-wide and market-segmented data and analytics.
Financial institutions can now access platforms and tools that help identify opportunities for margin improvement and quantify the impact of pricing decisions on new deposit growth, retention of existing balances, and customer acquisition. They can apply facts and data to the perpetual challenge of balancing pricing and volume — achieving the targeted growth needed to meet funding objectives that maximizes net interest margin.
This is achieved by employing constrained optimization modeling of pricing elasticity and marginal cost of funds. This leverages advanced machine learning to capture variation by product, geographic region, and customer segment. That analysis then rolls up to overall portfolio performance.
In the rising rate environment that began in March 2022, banks that used this kind of optimization grew deposits more than their peers. And their cost advantage expanded to 15 basis points by Q2 2023.
However, the limitation of today’s state of the art is “time to action.” This results from bottlenecks in efficiency and accessibility.
Typically, the end users in a financial institution are pricing analytics teams. They run a multitude of scenario models. Taking the optimal scenario to market testing generally takes weeks. The results of each optimization scenario generally need to be interpreted by that same team of specialized analysts and then communicated to business decision makers to determine whether to operationalize the pricing.
This requires organization of the findings for product, treasury, ALCO and marketing executives in a manner that is straightforward to understand and act on. But all that takes time and effort.
While industry results demonstrate that the outcome is demonstrably better than rate surveys and intuition, the time involved can result in missed market opportunity. In other instances, an institution won’t have the resident expertise required to go through these steps, nor the budget to acquire it.
Enter generative AI.
Getting Pricing Strategy in Place While It Still Matters
When applied to deposit-optimizing technology, generative AI can significantly accelerate decision-making and time to market for deposit pricing refinements. This reduces the effort needed to interpret results, and present findings in straightforward language with supporting data assets that virtually any responsible party in the organization can act on.
Such accessibility can allow the results of scenario-based queries to be delivered to a decision-making audience within hours. Rather than requiring a team of analysts to pore over and report on those results, the scenario output can be organized in a manner that generally requires the attention of only one specialist to audit and validate.
That individual can then directly present the results to management, complete with charts, tables and reports, with minimal manual intervention or formatting. In many cases, the responsible managers themselves can interpret and act on the output with no need for a human intermediary. This can accelerate the decision-making process and the impact of launching a given pricing approach.
Even though it’s still early days, it’s safe to say that AI’s impact on managing deposit pricing is a sea change in the making.
Read more:
Keeping Your Bank Out of AI-Generated Trouble
But as with any emerging technology that has the potential to transform, GenAI also presents potential hazards.
When applying AI to deposit pricing and in the banking context in general two features are paramount. First, the capability must be accurate, with fact-based output that can stand up to audits. Second, it needs to be compliant.
The outputs of any AI functionality supporting deposit pricing and scenario modeling need to be based purely on facts and immune to the hallucinations commonly experienced in general-purpose, publicly available large language models.
That requires a multi-agent approach that uses agent-experts to interact with accurate underlying data and deterministic pricing algorithm input parameters and outputs.
Such an approach employs separate agents to constrain the scope of application to a descriptive interpretation of the deposit pricing domain, and to a narrow dictionary of industry-specific terms and concepts. Such a constraint needs care to achieve, and requires an underlying software and data system structure that exposes reliable fact-based sources to each agent.
These underlying algorithms and the APIs rely on the same science by which scenarios are modeled by an analyst team. This allows consistent understanding and stable methodology from the perspective of an institution’s risk-management functions.
As with any other form of automation, AI’s outputs are only as good as its inputs, despite its game-changing sophistication.
The other critical component of any such AI-fueled platform is compliance with banking regulation.
Ideally, that means systems that are built from the ground up exclusively using banking industry principles that emphasize auditability of decision-making, strict mitigation of disparate impact, and strong data privacy and security protections.
That has multiple impacts on how the technology is used. First, it precludes the use of models that are hosted by third parties. Second, the institution’s data shouldn’t be made available for GenAI learning. Data on an institution’s product structure and balance sheet components must never leave the system’s data infrastructure.
Read more: Rethink Early Withdrawal Penalties to Create a Win-Win for Banks and Their Customers
How GenAI-Based Deposit Pricing Could Work in Your Bank
Let’s look at a concrete example. A deposit pricing manager at a large regional bank might be tasked by the head of retail banking to model a pricing optimization to grow money market account balances by 70% while minimizing overall interest expense — perhaps to fund the bank’s expected lending demand.
A typical optimized rate grid might contain tens of thousands of pricing cells — combinations of product features and customer attributes such as geography, balance tier and depth of relationship with the bank.
The output to the deposit pricing manager would look like pricing, margins and expected balances across those numerous cells, requiring significant further analysis and distillation to provide to the head of retail to put into effect as the bank’s product offerings. But with AI, the output can be executive- and execution-ready:
Leaders can see an approach set up in this way, and producing results like these:
“Optimization grew promotional MMA deposits to $10.95 billion by pricing at 4.05% on average. The product is projected to grow by $4.1 billion over the course of the scenario to hit this target.
Rates were set at 1.50% for the lowest balance tier and locked together across the other balance tiers. Rates between 200 and 450 basis points were tested.
• Kentucky and Indiana experienced the largest balance increase by far.
• Kentucky grew about $670 million by pricing at 4.49%, almost hitting the upper bound of 4.50%.
• Indiana grew by $574 million by pricing at 3.81%.
• Three markets grew between $280 million and $320 million:
• Raleigh grew by $320 million by pricing at 4.48%.
• Atlanta grew by $293 million by pricing at 3.77%.
• Louisville grew by $288 million by pricing at 4.24%.
Generative AI is already transforming what it means to price and manage deposits for optimal efficiency and profitability. It democratizes the power of pricing analytics across a wide range of users, from analysts and product managers to senior executives. For anyone responsible for optimizing funding costs, its future is now.
