Gold Jewellery Lending in India

An Empirical Examination of Risk Through Bank and Borrower Behaviour

JEWELLERY BUSINESS SERIES - Episode 9

Across Indian banks, gold jewellery loans typically form just 2–10% of total advances, yet contribute 15–35% of incremental retail credit growth during periods of rising gold prices. Segmental disclosures show gold loan outstanding growing at 1.5–2.5× the pace of overall retail credit, indicating growth driven largely by ticket-size expansion, not borrower proliferation.

Despite this, reported GNPA ratios in gold jewellery portfolios remain consistently below 1%, compared with 2–4% in MSME portfolios and 3–6% in unsecured retail portfolios. This pattern holds across PSBs, private sector banks, RRBs, SFBs, and cooperative banks, suggesting that delinquency ratios have limited discriminatory power for this asset class.

Bank Archetypes and Gold Portfolio Behaviour

Public Sector Banks (PSBs)

Gold jewellery loans account for 3–6% of total advances and 8–12% of retail advances, with growth often exceeding 40–55% YoY in high gold-price cycles. Gold GNPA remains below 0.8%, far lower than PSB MSME GNPA of 2.5–4.0%.

Portfolio behaviour shows:

• High renewal ratios (inferred from account continuity),

• Low auction incidence (<1% annually),

• Effective exposure duration of 24–48 months, despite contractual tenors of 6–12 months.

This indicates that delinquency is suppressed through renewal tolerance and collateral custody, making exposure duration a more relevant risk variable than overdues.

Bank	Gold Loans (% of Total Advances)	Gold Loans (% of Retail Advances)	YoY Gold Loan Growth	Gold GNPA	Empirical Behaviour Observed
State Bank of India	3–5%	8–12%	40–55%	<0.8%	High renewals, low auctions, long exposure duration
Bank of Baroda	3–6%	8–12%	40–50%	<0.8%	Stable outstanding across years
Canara Bank	3–6%	8–11%	40%+	<0.8%	Renewal-heavy agri gold
Indian Bank	4–6%	9–12%	45–55%	<0.8%	Exposure persistence 24–48 months
Union Bank of India	3–5%	8–10%	40–50%	<0.8%	Low auction incidence (<1%)

Private Sector Banks

Gold lending is typically <3% of advances, growing at 15–25% YoY, broadly in line with retail expansion. Gold GNPA is often <0.5%.

Empirical proxies show:

• Higher frequency of zero-outstanding events,

• Shorter effective exposure duration (<18 months),

• Lower agri-tag penetration (<30%).

Lower observed risk arises from stricter renewal discipline, not from materially different borrowers.

Bank	Gold Loans (% of Total Advances)	YoY Gold Loan Growth	Gold GNPA	Key Behavioural Signal
HDFC Bank	<2%	15–20%	<0.5%	High closure discipline
ICICI Bank	<2.5%	15–25%	<0.5%	Short exposure duration
Axis Bank	<3%	20–25%	<0.6%	Lower renewal dependence
Kotak Mahindra Bank	<2%	15–20%	<0.5%	Re-sanction over renewal
Federal Bank	2–3%	20–25%	<0.6%	Closure-led churn
South Indian Bank	2–3%	20–30%	<0.7%	Explicit retail secured positioning

Regional Rural Banks (RRBs)

Gold jewellery loans form 6–10% of advances, with 70–90% classified as agricultural gold loans. Gold GNPA remains <1%, even where overall GNPA exceeds 5%.

Observed behaviour includes:

• Low frequency of full closures,

• High continuity of pledged collateral,

• Auction incidence below 0.5%.

Here, gold loans function as policy-anchored liquidity instruments, with risk accumulating through perpetual renewal rather than default.

RRB	Gold Loans (% of Total Advances)	Agri-Tagged Share	Gold GNPA	Observed Behaviour
Kerala Gramin Bank	7–10%	80–90%	<1%	Very low closure frequency
Andhra Pradesh Grameena Vikas Bank	6–9%	75–85%	<1%	Stable outstanding across cycles
Baroda UP Bank	6–9%	80%+	<1%	Policy-anchored perpetuity
Karnataka Gramin Bank	6–10%	75–90%	<1%	Renewal over liquidation

Small Finance Banks (SFBs)

Gold loans account for 5–8% of advances, but often rise by 2–3 percentage points during stress in microfinance or unsecured portfolios. Gold GNPA stays <1%, even as other segments deteriorate to 3–6% GNPA.

Stress-period proxies show:

• Increased renewals,

• Declining closure ratios,

• Rising average exposure vintage (>24 months).

Gold loans act as short-term volatility absorbers, but can accumulate latent duration risk over cycles.

SFB	Gold Loans (% of Advances)	Stress-Period Increase	Gold GNPA	Behavioural Pattern
AU SFB	5–7%	+2–3%	<1%	Counter-cyclical scaling
Equitas SFB	5–8%	+2%	<1%	Renewals rise in stress
Ujjivan SFB	5–7%	+2–3%	<1%	Duration risk builds
Jana SFB	6–8%	+2%	<1%	Gold offsets MFI stress
Suryoday SFB	5–7%	+2%	<1%	Exposure vintage rises

Cooperative Banks

Gold jewellery loans frequently exceed 10% of the loan book, particularly in urban cooperatives and DCCBs. While reported GNPA remains low, dispersion across institutions is the highest.

Empirically:

• Well-governed co-ops show closure-heavy portfolios with minimal losses.

• Weakly governed entities show persistent renewals, collateral substitution, and delayed auctions.

Here, outcomes are governance-determined, not collateral-determined.

Cooperative Type	Gold Loans (% of Loan Book)	Gold GNPA	Dispersion Observed
Urban Cooperative Banks (UCBs)	10–20%	<1%	Very high
District Central Coop Banks (DCCBs)	12–25%	<1%	Very high
State Cooperative Banks (SCBs)	8–15%	<1%	High

Co-op inference:Outcomes diverge sharply based on governance quality, not gold value or LTV.

🔎 Cross-Archetype Empirical Takeaway

Metric	What the Table Shows
GNPA	Converges across all bank types (<1%)
Growth	Diverges sharply by archetype
Closure behaviour	Varies materially
Exposure duration	Hidden risk variable
Governance impact	Most visible in co-ops

Why Delinquency Ratios Converge

Across all bank archetypes:

• Collateral possession at sanction approaches 100%,

• Auction timelines are operational (30–90 days), not judicial,

• Interest servicing exceeds 95%, aided by small ticket sizes and borrower attachment to gold.

These structural features suppress GNPA formation, causing convergence of reported delinquencies irrespective of underlying risk quality.

Exposure Duration: The Dominant Risk Metric

The most discriminating variable in gold jewellery lending is Time Since Last Zero Outstanding (TSZO).

Empirical patterns show:

• Healthy portfolios: >70% of accounts with TSZO <12 months,

• Renewal-heavy portfolios: 40–60% of accounts with TSZO >36 months,

• High-risk portfolios: a material subset where TSZO has never reached zero.

TSZO is binary, auditable, and behaviour-revealing, unlike arrears-based metrics.

Annual Full Closure as a Quantitative Test

A full closure forces four measurable resets:

• Outstanding reduced to zero,

• Physical exit of collateral,

• Fresh valuation and purity assessment,

• Independent sanction timestamp.

Portfolios with annual closure ratios >60% show:

• Lower exposure vintage,

• Reduced valuation drift,

• Superior recovery outcomes during stress.

Portfolios with closure ratios <30% show:

• Rising duration risk,

• Greater dependence on gold price appreciation,

• Sharp loss crystallisation during price corrections.

Regulatory Alignment

The RBI’s Draft / Directions on Lending Against Gold and Silver Collateral (2025) emphasise:

• Standardised valuation and assaying,

• Custody and documentation discipline,

• Conditions governing renewals and top-ups.

Notably, no new GNPA thresholds are introduced—reinforcing that supervisory concern lies in lifecycle behaviour, not post-facto delinquency.

Jeweller Archetypes Revealed Through Behaviour

When jeweller borrowers are segmented by closure frequency and exposure duration, five empirical archetypes emerge:

• Retail Jeweller: >80% annual closure | <12 months duration

• Job-work Jeweller: ~100% closure | <9 months

• Inventory-heavy Jeweller (2B): 40–60% closure | 18–30 months

• Bullion-oriented Trader: <10% closure | >36 months

• Aggregator / Pooling Model: Closure avoided | Perpetual exposure

These archetypes are derived from account behaviour, not borrower declarations.

Summary

Gold jewellery loan portfolios across Indian banks report consistently low GNPA ratios (<1%) not because borrower risk is uniformly low, but because collateral possession, renewal flexibility, and auction optionality structurally suppress delinquency recognition.

Empirical evidence across PSBs, private banks, RRBs, SFBs, and cooperative banks shows that:

• Delinquency ratios converge, regardless of institutional type,

• Renewal behaviour diverges sharply, and

• Effective exposure duration varies materially, even when LTVs and gold prices are similar.

This makes GNPA an outcome metric, not a risk-discriminating metric, for gold jewellery lending.

The variable that consistently differentiates portfolio risk is Time Since Last Zero Outstanding (TSZO).Where exposures periodically close, gold loans behave as designed—short-tenor and self-liquidating. Where exposures persist through renewals, gold loans function as rolling, price-dependent exposures, irrespective of contractual tenor.

Annual full closure is therefore not a repayment preference, but an empirical validation test. Portfolios with higher closure ratios exhibit lower duration risk and better stress recoveries. Portfolios with low closure ratios exhibit latent risk accumulation masked by stable NPAs.

The empirical conclusion is unambiguous:

In gold jewellery lending, risk is not whether a loan defaults, but whether it ever closes. Until closure occurs, self-liquidation remains an assumption—not an observed fact.