Managed Rebound

The taka's September comeback has a problem. It happened while the dollar was getting stronger, not weaker. On September 16 the Federal Reserve raised rates to 3.75-4.00 percent, the dollar index broke above 100, and the rupee and the rupiah fell. The taka went the other way: from Tk 123.95 on August 30 to Tk 123.00 on September 17, and Tk 122.70 by September 21. A currency does not firm into a strengthening dollar by accident. Something domestic is doing the work.

That something is mostly Bangladesh Bank. The central bank bought dollars from commercial banks in early September, its first such purchase in three months, because, in the press's words, an abundance of foreign currency threatened an abrupt drop in the dollar's value. Read that again: the bank intervened to stop the taka from rising too fast. The recovery the press celebrates is a rally the central bank is actively capping.

The arithmetic, first

Before the story, the corrections. A Bangla Tribune report this week says the taka "lost 45.05% of its value" as the dollar rose from Tk 85.80 to Tk 123.95. That is the wrong number. The dollar's taka price rose about 44.5 percent. The taka's actual loss of value is one minus (85.80/123.95), roughly 31 percent. Reporting the dollar's appreciation as the taka's depreciation is a standard press error, and it matters here because the whole recovery narrative is built on percentages that were never computed correctly.

The recovery evidence is thin on its own terms. Tk 123.95 to Tk 123.00 over eighteen days is a 0.77 percent move. That is stabilization, not a reversal. The "0.06 percent gain over the year to June 2026" cited in the same report is noise dressed up as data. And the claim that the Sri Lankan rupee weakened alongside the Indian rupee and Indonesian rupiah is stale: over the September 17 to 23 week, the Sri Lankan rupee actually firmed about 0.8 percent, from 331.76 to 328.96 per dollar, tracking Bangladesh more than India.

The anomaly

September 16, 2026: the Federal Open Market Committee voted 12-0 to raise rates 25 basis points to 3.75-4.00 percent, the first hike since July 2023, under new Chair Kevin Warsh, and signaled one more hike before year-end. August producer prices printed 5.4 percent year on year. The 10-year Treasury topped 5 percent. The dollar index broke 100.00 the next day, a seven-week high, and kept climbing. Every peer logic says the taka should have weakened. It did not.

The magnitude is small. The direction, against a rising dollar, is the story. The drivers are domestic plumbing, not macro: remittance formalization, a hundi crackdown, and a central bank leaning against the rally it claims to welcome.

Four claims, tested

Four explanations are circulating for the taka's firming, including one supplied to this newsroom. Each was tested against the documents. Verdicts are Confirmed, Contradicted, or Unsupported.

The managed rate

Bangladesh runs a crawling peg, introduced May 8, 2025. The IMF's Article IV consultation report, released February 27, 2026, is blunt about how it operates in practice: the bank "either did not announce intervention volumes or failed to adhere to those announced" and "decided actual purchases to keep the dollar-taka rate near a nearly flat, preferred level." The government's position to the Fund is that the rate has been fully market-determined since May 2025. The Fund's report says otherwise, on the record.

The bank's own spokesperson, Arif Hossain Khan, told BSS on September 8 that Bangladesh Bank "has also purchased dollars from the domestic foreign exchange market when market conditions allowed," helping rebuild reserves. Gross reserves stood at $36.27 billion on September 10 ($31.36 billion on the IMF's BPM6 measure), down about $1.1 to 1.2 billion from end-June after a $1.39 billion payment of import and other bills in early September. The IMF's net-reserves target was met. The reserve recovery, a CPD economist notes, coincides with private-sector credit growth below 5 percent and falling productive imports: capital-machinery LC settlements down 10.5 percent, intermediate goods down 6.5 percent in FY26. Reserves rebuilt while the real economy borrows less is not obviously a vote of confidence.

The street disagrees

There are two dollar prices in Dhaka, and they tell two stories. Around September 22, the interbank spot rate sat at Tk 123.52. The open market in Motijheel was selling at Tk 126.50 and buying near Tk 126.20, a premium of about 2.4 percent. A week earlier the street was at Tk 125.00. The official rate firmed; the street rate rose. One Motijheel trader put it plainly: there is simply not enough supply to meet market demand, which is why prices are increasing.

The kerb premium is the market's vote on the official rate. As long as it persists at this width, the screen understates true depreciation pressure, and the 0.77 percent "recovery" is a number on a managed board, not a price the street will trade at.

The plumbing underneath

The genuine good news is remittances. FY26 brought a record $35.34 billion through formal banking channels. July 1 to September 20 of FY27 brought $7.82 billion, up 14.9 percent year on year; the first twenty days of September alone brought $1.99 billion, up 4.8 percent. The 2.5 percent cash incentive remains in place, the official-street spread has narrowed enough to reduce hundi's edge, and the crackdown on informal channels is pushing flows into banks. As one bank CEO put it, with the incentive on the table, sending through hundi is now a loss for the remitter.

Note the ambiguity inside the good number. Part of the "surge" is formalization: dollars that previously traveled through hundi now travel through banks. Measured inflows rise without a single new dollar entering the economy. That is still an improvement, the central bank sees the flows and the reserves, but it is not the same as the diaspora sending more.

The import side is liberalizing, not compressing. The Import Policy Order 2026-29, gazetted August 24, scraps the mandatory letter of credit for many imports and removes the old $500,000 annual non-LC ceiling. FY26 LC settlements were $70.41 billion, essentially flat year on year. The cabinet has approved an extra 695,000 metric tonnes of fuel oil under government-to-government deals for September to December, 10 percent above the approved 2026 quantity. None of this is demand suppression. If anything, the import bill has room to grow into the dollar surplus and erase it.

What to watch

Six observables, stated so they can be proven wrong. First, the kerb premium: about 2.4 percent on September 22. If it narrows toward zero, the official rate is gaining credibility. If it widens, the screen is fiction. Second, the bank's dollar purchases: continued buying confirms the ceiling on appreciation; a sustained pause lets the market speak. Third, September EPB export data, not yet published: does the garment surge extend past August, or was it a tariff-settlement bounce? Fourth, the sixth IMF tranche: undisbursed as of September 23, its release or continued delay. Fifth, the Rooppur $630 million: a sanctions-blocked payment channel with a passed deadline. Sixth, reserves: the $1.1 to 1.2 billion quarterly decline against the remittance surge, and whether the next ACU-cycle payment knocks them again.

The kill condition for this article's thesis: if the kerb spread collapses toward zero and the bank stops intervening in either direction for a sustained stretch, the "managed" in the headline fails, and the rebound was real.