Trinidad Held Its Repo at 3.5%, but Credit Is Slowing

By Lancelot Williams | October 12, 2026

The Central Bank of Trinidad and Tobago kept its repo rate at 3.50 per cent on 30 September 2026, and the same announcement reported that consolidated system credit growth had slowed to 3.4 per cent in July, with business lending up only 0.6 per cent on the year. The policy rate did not move and inflation is 0.6 per cent, yet credit is slowing. For a small business owner, the practical job before 2 November is to build the sales record a lender can actually read.

The Short Version

What Happened at the Central Bank on 30 September

In its monetary policy announcement of 30 September 2026 (full text in the PDF), the Monetary Policy Committee decided to maintain the repo rate at 3.50 per cent. Its reasons were domestic: "the moderation of economic activity, slowing private sector credit expansion, and low inflation conditions". The next announcement is scheduled for 23 December 2026.

The outside world looks different. The Bank notes that central banks in the US, Japan and the euro area have adjusted their policy rates upwards, and quotes the IMF's July projection of global growth at 3.0 per cent for 2026 and 3.4 per cent for 2027, which the IMF update confirms. At home, the Central Statistical Office put headline inflation at 0.6 per cent for July 2026. Low headline inflation and slower credit growth appear together in these figures. Neither establishes why a particular business was refused a loan.

The Numbers in One Place

The first five figures come from the Central Bank's 30 September announcement. The last two come from the survey that the Trinidad and Tobago International Financial Centre (TTIFC) ran with the UN Capital Development Fund, in the survey report and the TTIFC's overview of its findings.

What the Rate Leaves Out of a Loan Application

The repo rate is a central bank policy rate, not the rate every borrower pays. For a hairdresser in Arima or a caterer in Chaguanas, a lender will also assess repayment capacity and its own lending requirements. The financial inclusion survey describes some obstacles to building that relationship. It sampled 2,000 people and found that 25 per cent of adults lacked a basic transaction account. On account ownership, the report's own comparison puts the 2023 figure of 75 per cent below the 81 per cent recorded in 2017. Among responding MSMEs, 77 per cent had no business bank account and only 10 per cent accepted digital payments linked to one.

The costs are written into the report. The 2024 report lists merchant terminal fees of TT$300 to TT$600 a month and card transaction fees of 1.5 per cent to 5 per cent. These are historical ranges, not current quotes. For a stall selling low-priced items one customer at a time, those fees are hard to absorb. We say so as a reading of the fee table, not as a finding of the survey. Among the financially excluded, 41 per cent said they lack the funds to register or use an account, and 13 per cent named documentation.

A cash business may still have invoices, tax records, contracts or a repayment history. A missing bank statement does not mean a business has no evidence. Our suggestion is to organise the records it does have and ask a lender what else is needed. These sources do not tell us why national business credit growth slowed, or how much missing documentation contributed.

The Gap Is Also About Know-How

Access is not the whole story. The TTIFC overview says 56 per cent of account holders lack the knowledge to use mobile banking apps, and describes general information from the financial sector as often seen as "cumbersome, confusing, and, at times, hostile". Distance plays a part too: only 46 per cent of respondents live within a 15-minute radius of a bank branch or ATM, which is where the phone could help if people knew how to use the app.

The overview also groups car loans, mortgages, crypto-assets, bonds, stocks and investment funds into a broad category held by 3 per cent of respondents. That combined measure is not a borrowing rate or a measure of creditworthiness. It cannot tell a lender whether a particular owner can repay. An individual application needs evidence about that business, its costs and its existing obligations.

Applying "People First, Problem Second, AI Last"

This is a method Adrian Dunkley has published on his FAQ page (adriandunkley.net): start with the human need, define the problem precisely enough to know what a solution would look like, then decide whether AI is the right tool. Using it on small business credit is our extension of the method, not something he has published about Trinidad and Tobago.

Take a hypothetical caterer who wants a second freezer before the Christmas season. The need, written without the word AI: she must hold enough stock to fill December orders. The problem: she is paid in cash and has no dated sales record to show. She can begin with an exercise book, then ask the lender which other records it accepts. The log will not settle questions about costs, existing debt or whether she can afford the freezer.

AI might later help categorise records or flag entries for checking. Turning those records into a credit score requires separate validation, lawful data use and a lender willing to accept the result. Three months of sales alone cannot establish a model's reliability. Credit Garden describes its World Credit Score as reflecting country conditions, including inflation, GDP, minimum wage, unemployment and financial inclusion. That is the company's description, not independent evidence that the score predicts repayment or is accepted by Trinidad and Tobago lenders.

What This Means in Trinidad and Tobago

Put the Central Bank's picture beside the survey and the pressure points are specific. Unemployment rose to 5.4 per cent in the first quarter of 2026 from 4.9 per cent a year earlier, and the Bank says labour demand softened through July. Consumer lending was still growing at 4.3 per cent, while business lending growth was 0.6 per cent. These are aggregate balances, not approval rates: they do not show which households or firms could obtain a new loan.

Compare Jamaica, where our last article covered the Bank of Jamaica lifting its policy rate to 6.0 per cent with inflation at 7.9 per cent. Read it here: Bank of Jamaica lifts its rate to 6.0%. The two policy settings differ. Neither policy rate tells an owner what a lender will charge or whether an application will succeed. Our general guide to credit and finance in Trinidad and Tobago covers the basics. For the regional payments side, see our CAPSS piece.

Where AI Fits, and Where It Does Not

A lender could test whether verified sales and supplier-payment records add useful information to its existing assessment. The test must include cash businesses and check errors, fairness and affordability before the model influences a decision. A model cannot verify a sale merely because someone typed it into a spreadsheet. The survey's 86 per cent digital-payment figure is a reason to retain ways of assessing cash records, not a reason to assume those businesses are poor credit risks.

Risks to Weigh

The main risk is a model trained on salaried, bank-account data that scores a cash trader as risky because it never sees her sales. That possibility should be tested across home-based firms and different locations; these sources do not establish which group has the thinnest records or the highest model error rate. A second risk is false confidence: a score that looks precise but rests on three months of data. A third is privacy, since payment records shared with a lender or a platform are personal. The guardrails are plain. Ask which data a lender uses, how to correct an error and whether a human can review a refusal. Use the lender's verified secure channel and ask about access, retention and deletion. Do not upload customer details to a public AI chatbot.

Do This Week

1. Start a dated sales log today. One line per sale: date, amount, how it was paid. A notebook is enough.
2. Ask your bank for its business account requirements in writing. The survey says 13 per cent of the financially excluded name documentation as the main barrier, so find out which documents you are missing.
3. Offer one traceable way to pay. A bank transfer, card or wallet, whichever your customers already use. Ask for current setup, monthly and transaction fees in writing. Compare the total cost with your likely sales; the survey's TT$300 to TT$600 terminal range is historical.
4. A test we suggest, with a deadline. From Tuesday 13 October to Monday 2 November 2026 (21 days), log every sale. At the end, count the sales recorded, reconcile daily totals with receipts or cash counts, and note missing entries and the time spent. Report the share paid by each method without treating cash as a failure. A lower-volume business can still learn whether its records are usable. This exercise cannot predict loan approval.
5. Then check your profile. Bring the log to your lender and ask what additional evidence it needs. The World Credit Score calculator is a separate Credit Garden tool; do not treat its output as a lender's decision or a guarantee of credit.

A Loose End

The weakest part of this argument is the bridge from the survey to the credit slowdown. The survey is from 2023 and covers household-level MSMEs, and the Central Bank's credit figures are year-on-year growth rates for the whole system. I found no source that says small firms are the reason business credit slowed, and the Bank does not say what drove the 0.6 per cent. If you have a lender's view on why, I would like to hear it. I also did not find a recent update to the 77 per cent figure.