Nu Holdings: Protecting Credit Quality Comes at an Earnings Cost
Tags: Nu Holdings; Brazil consumer credit; earnings revisions; underwriting; valuation; macroeconomic risk
Value Score: 8/10
Core Conclusion: Morgan Stanley retains Overweight on Nu Holdings, but its November 10, 2026 forecast update makes the near-term trade-off explicit: tighter underwriting can protect credit quality while quickly reducing income from a short-duration loan book. The bank cuts 2027 and 2028 EPS estimates by 24% and 27%, to US$0.99 and US$1.29, and lowers its year-end 2027 price target to US$17 from US$21. Brazil's employment outlook and Nu's lending response determine whether the reduced forecasts are sufficient; technology advantages and international expansion support the longer-term case.
Slower Lending Drives the Earnings Reset
The central earnings risk is weaker origination, rather than an uncontrolled increase in delinquencies. Morgan Stanley expects Nu's data and underwriting systems to identify consumer stress early and restrict new lending. Because existing loans mature relatively quickly, reduced origination can rapidly weaken net interest income. Expenses adjust more slowly, amplifying the effect on profit. This is the analyst's forecast mechanism under a tougher Brazilian outlook, not a measured post-election outcome.
For 2027, the bank lowers Brazilian credit-card receivables growth to 19% from 26%, personal-loan growth to 17.5% from 24.5%, and secured-loan growth to 41.5% from 84.4%, all at constant exchange rates. The detailed commentary supplies the more precise personal- and secured-loan assumptions; Exhibit 1 rounds those figures. Consolidated interest income falls to US$23.4 billion in 2027 and US$28.2 billion in 2028, respectively 8% and 11% below the previous forecasts. These figures measure interest income before funding expenses, rather than net interest income.
Higher credit-loss rates coexist with lower total provisions because the expected lending book is smaller. The 2027 Brazilian card cost of risk rises to 14.5% from 13.6%, while personal-loan cost of risk rises to 28.6% from 27.1%. Yet consolidated loan-loss provisions decline by 8% to US$8.3 billion in 2027 and by 15% to US$9.5 billion in 2028 relative to the previous model. Lower provisioning dollars therefore do not signal an improved borrower-risk assumption.
The reduction in expenses only pa