Bangladesh Policy Bundle Digital Twin: A Five Layer Simulation Case Study

Why a Digital Twin for the Graduation Bundle #

Bangladesh has cleared the United Nations Committee for Development Policy thresholds on income, human assets, and economic vulnerability for two consecutive triennial reviews and is scheduled to graduate from least developed country status on 24 November 2026. The trade preference cliff is real: the European Union Everything But Arms regime ends after a three year transition, the United Kingdom Developing Countries Trading Scheme tier shifts, and most other partners convert duty free access into standard Generalized System of Preferences with rules of origin that bind ready made garment exports more tightly than today.

Cabinet level decisions in this window cannot be reduced to a single elasticity. They sit inside an institutional graph of ministries, standing committees, autonomous regulators, and donor counterparts; they touch four or five policy domains simultaneously; and they are constrained by overlapping legal frameworks from the Customs Act to the Bangladesh Bank Order. A digital twin that respects all five of these layers, rather than collapsing them into a reduced form macro model, is the difference between a memo a secretary can sign and a forecast a planner discards.

The Aegis platform is our internal scaffold for that work. It is not a forecasting tool in the conventional sense. It is a simulation environment that lets a small policy team replay a shock through the actual administrative geometry of the state, with named decision rights at each node.

The Five Layer Architecture #

Each layer is a graph or schema with explicit edges to the next. The institutional graph names the actors and their formal authorities; the policy domain layer groups instruments by economic function; the legal framework layer pins each instrument to the statute or rule that authorizes it; the committee layer records who must concur, in what sequence, before an instrument moves; and the peer country layer holds calibrated analogues from comparable transitions.

Building the layers is roughly two thirds of the total effort on a country engagement. Once they exist, swapping in a new shock takes days rather than months, which is the methodological payoff. The table below summarizes how the layers were instantiated for the graduation bundle.

The Shock: GSP Loss, EBA Cliff, Debt, RMG Competitiveness #

The shock bundle has four coupled components. First, loss of EBA after the November 2029 transition window, which restores a most favored nation tariff averaging roughly 9.6 percent on woven garments and 11.5 percent on knitwear into the European Union, currently the destination for around 54 percent of ready made garment exports. Second, a parallel loss of duty free access in Canada, Japan, and Australia on staggered timelines. Third, a tightening external financing environment in which International Development Association concessional windows narrow and the country shifts toward International Bank for Reconstruction and Development pricing, raising the implicit cost of the existing public and publicly guaranteed debt stock. Fourth, a competitiveness layer in which Vietnam, Cambodia, and Pakistan continue to gain share through free trade agreements and lower unit labor cost growth.

Inside the twin, each component is encoded as a parameter change at a specific node, not as a free floating macro shock. The EBA cliff is an edge weight change on the trade domain node connected to the European Commission counterpart. The IDA to IBRD transition is a schedule change on the external debt sub graph owned by the Economic Relations Division. This discipline matters because it forces the modeling team to identify, before any simulation runs, which committee owns the response at which node.

Propagation Across Budget, Balance of Payments, and Labor #

Aegis propagates a shock through three coupled accounts in sequence and then iterates to fixed point. The budget account captures revenue, current expenditure, and the financing gap; the balance of payments account captures the current account, capital flows, and reserve adequacy; the labor account captures formal sector employment, informal absorption, and remittance feedback. Each pass updates the others until the deltas across an iteration fall below a tolerance.

For the central calibration, a full EBA loss with no policy response shifts the goods trade balance by roughly 1.7 to 2.4 percent of gross domestic product within three years of the cliff, with a corresponding pressure on gross reserves of between 4 and 6 weeks of imports depending on remittance behavior. Revenue effects are second order in the first year because import compression partially offsets lost preference rents on inputs, but they widen materially by year three as ready made garment value added contracts. The labor account shows a slower adjustment: roughly 380,000 to 520,000 fewer formal ready made garment jobs by year three relative to the no shock counterfactual, concentrated in Gazipur, Narayanganj, and Chattogram, with informal sector absorption visible but incomplete.

These are not point forecasts. They are the central path of an ensemble in which the dispersion is driven by remittance elasticity, the speed of the European Union response on rules of origin cumulation, and domestic pass through to wages. The twin reports the ensemble; the cabinet committee picks a planning case.

Three Named Scenarios #

The engagement specified three scenarios by name, each tied to a different policy posture. Padma Bridge holds existing instruments roughly constant and absorbs the shock. Jamuna Pivot pursues an aggressive trade diversification and free trade agreement agenda alongside a managed taka depreciation. Meghna Compact pairs domestic tax reform and a social protection expansion with negotiated transition arrangements with major partners.

Each scenario is a bundle of instrument settings across the policy domain layer, with explicit committee approval pathways encoded. The simulation does not assume frictionless implementation: each instrument carries a probability of timely passage based on the legal framework and committee load. The table below summarizes the three year outcomes for the central calibration.

What the Twin Changes in the Room #

The most useful artifact in cabinet committee discussion was not any single number. It was the ability to point at a node, ask what happens if this circular is delayed by ninety days, and see the propagation in budget and reserves within minutes. That collapsed a class of arguments that had previously consumed entire meetings, in which different secretaries held different mental models of the same instrument.

The peer country layer also did unexpected work. Vietnam in 2007 is not a template for Bangladesh in 2026, but the calibrated analogue forced participants to be specific about why their case differs. Cabo Verde and Samoa, both small, supplied the discipline of bounding optimism on tourism and remittance substitution. The Maldives case anchored discussion of the debt sustainability tail.

Limits matter too. The twin does not model political economy beyond committee load and instrument probabilities. It does not capture climate shocks endogenously, though it carries them as overlay scenarios. And it is only as good as the institutional graph, which decays as ministries reorganize. We refresh the graph quarterly under the engagement.

Methodological Takeaways for Other Engagements #

Three points generalize beyond Bangladesh. First, the value of a digital twin in government work comes from respecting administrative geometry, not from a more elaborate macro core. A standard small open economy model wrapped in a careful institutional graph beats a fancy DSGE without one. Second, naming scenarios after recognizable national landmarks, rivers, or programs improves uptake; abstract labels like A, B, C lose the room. Third, the layers must be owned by the client, not by the consultancy, or the artifact dies at handover.

Aegis is the anchor we use to enforce these disciplines across engagements, from graduation transitions in South Asia to fiscal rule design in Latin America. The Bangladesh case is the clearest recent demonstration because the shock is dated, the institutions are well documented, and the policy response window is short enough that the twin paid back its build cost inside a single budget cycle.

Sources #

• Bangladesh Bank statistics
• Bangladesh Bureau of Statistics
• UN LDC Portal Bangladesh graduation
• IMF Article IV Bangladesh
• World Bank Bangladesh Development Update