Skip to content
TheChatArchiveThe Chat Archive
Grok 4.1 Fast· xaiHugging Face Frontier Reasoning Traces

Design a treasury FTP for deposits: stickiness tiers, rate betas, deposit decay, FTP…

  1. User

    Design a treasury FTP for deposits: stickiness tiers, rate betas, deposit decay, FTP credits, and performance attribution.

  2. Assistant

    Treasury FTP Framework for Deposits

    As Sherlock, I've deduced the optimal design for a Funds Transfer Pricing (FTP) system tailored to deposits. This framework incentivizes stable, low-cost funding while accurately attributing performance to deposit teams. It balances stickiness (stability), rate sensitivity (beta), decay risk, FTP credits (internal pricing), and attribution for P&L analysis.

    The design assumes a Treasury curve (e.g., interpolated SOFR + tenor premium) as the base reference rate. FTP rates are calculated daily/weekly, with credits flowing from Treasury to deposit businesses.

    1. Stickiness Tiers

    Deposits are segmented into 4 tiers based on behavioral data (e.g., historical duration, volatility, relationship linkage). Higher stickiness = higher FTP credit (more value to the bank).

    TierDescriptionCriteriaStickiness ScoreFTP Premium (bps over base)
    CoreNon-interest bearing (NIB), operational cashAvg. life >5 yrs, volatility <5%100%+50 bps
    Sticky RelationshipCDs, MMDA linked to loans/relationshipsAvg. life 2-5 yrs, volatility <15%80%+30 bps
    Volatile RelationshipSavings, NOW accounts with some linkageAvg. life 1-2 yrs, volatility 15-30%50%+10 bps
    Hot MoneyBrokered CDs, time depositsAvg. life <1 yr, volatility >30%20%0 bps

    Assignment: Machine learning model (e.g., survival analysis) reclassifies quarterly based on 12-month rolling data.

    2. Rate Betas

    Beta measures deposit rate repricing sensitivity to market rates (e.g., Treasury/SOFR curve shifts). Estimated via regression on historical data.

    TierBeta (ΔDeposit Rate / ΔMarket Rate)RationaleAdjustment to FTP
    Core0.1Largely insensitive (corporate cash)-40 bps (low cost)
    Sticky Relationship0.4Partial pass-through-20 bps
    Volatile Relationship0.7High sensitivity0 bps
    Hot Money1.2Over-indexes market (competitive)+20 bps (penalty)

    Formula: FTP Rate = Reference Curve Rate × Beta + Stickiness Premium + Decay Adjustment.

    3. Deposit Decay

    Models expected balance runoff using exponential decay:
    Balance_t = Balance_0 × e^(-λt)
    Where λ (decay rate) varies by tier, calibrated to historical attrition.

    TierHalf-Life (t½)λ (Decay Rate, monthly)Decay Penalty (bps, cumulative)
    Core10 years0.007%-5 bps/yr
    Sticky Relationship4 years0.017%-15 bps/yr
    Volatile Relationship2 years0.035%-30 bps/yr
    Hot Money6 months0.116%-60 bps/yr

    Implementation: Forward-looking adjustment to FTP credit. E.g., for a $100M Volatile deposit: Expected decay over 1yr = 20%, so haircut FTP by 20% of margin.

    4. FTP Credits Calculation

    FTP Credit Rate = (Treasury Reference Rate × Beta) + Stickiness Premium + Decay Penalty + Volume Incentive.

    • Reference Rate: Weighted average cost of funds (WACF) from Treasury curve (e.g., 3M SOFR + 10bps credit spread).

    • Full Formula (annualized):

      FTP_Credit = [Ref_Rate × Beta] + Stickiness_Premium - Decay_Penalty + (Volume_Growth × 2bps)
      
    • Example (Ref_Rate = 5.00%):

      TierBeta Adj.StickinessDecayFTP CreditMargin to Deposit Business
      Core0.50%+50bps-5bps5.45%Deposit Rate (0%) → +545bps
      Hot Money6.00%0bps-60bps5.40%Deposit Rate (5.25%) → +15bps
    • Credits Flow: Monthly accrual: Credit = Avg Balance × FTP Credit Rate. Paid to P&L of originating unit.

    5. Performance Attribution

    Decomposes deposit P&L into actionable drivers for incentives/oversight. Uses marginal FTP (incremental credits on growth).

    Attribution Bucket% of Total P&LCalculationIncentive Weight
    Volume40%ΔBalance × Avg FTP RateHigh (growth rewarded)
    Mix Shift25%∑(Tier Weight × FTP Margin) - Prior MixHigh (stickier = better)
    Rate Capture20%(Market Δ × (1-Beta)) × BalanceMedium (low beta wins)
    Decay Control10%Actual Runoff vs. Modeled DecayHigh (retention bonus)
    Beta Stability5%Variance in Observed vs. Expected BetaLow (risk control)

    Dashboard Example:

    Deposit Team P&L: +$12M
    - Volume: +$5M (42%)
    - Mix: +$3M (25%) [↑Core 10%]
    - Rate: +$2.5M (21%) [Beta=0.3 vs. mkt]
    - Decay: +$1.2M (10%) [Runoff -5% vs. exp]
    - Beta: +$0.3M (2%)
    

    Implementation Notes

    • Tech Stack: Python (pandas for calc, scikit-learn for tiers/betas), integrated with core banking (e.g., FIS, Temenos).
    • Governance: Monthly calibration by ALCO; stress-tested for +200bps rate shock.
    • Incentives: Bonus pool = 20% of attributed margin > budget.
    • Edge Cases: New products default to "Hot Money"; grandfather legacy at tier average.

    This design promotes sticky, low-beta deposits while providing granular insights. If you supply bank-specific data (e.g., deposit book stats), I can refine with simulations. Elementary, my dear banker. 🔍

Responses were produced in different, uncontrolled contexts (system prompts, settings, prior turns). This is not a benchmark.