Most economic impact studies answer the first question: which industries gain or lose? The harder question is what happens next. When output changes, wages, profits, taxes, household income, consumption, and imports change with it. The effect moves through the economy rather than stopping at the sector where it began. Especially for a public-sector decision, a policy that looks positive in aggregate can still miss its target, shift losses to vulnerable groups, or fail when external conditions change.

M & Sullivan's SAM Model is our tailored implementation of the established Social Accounting Matrix methodology, bringing production, factor income, households, businesses, government, and the foreign sector into one balanced economic system. Instead of treating a policy as an isolated input, the model traces how its consequences circulate and accumulate.

The Limitation in Conventional Impact Analysis

Output analysis could trace direct and indirect effects between industries, but it could not fully capture the induced effects created when wages and profits become household income and return to the economy through consumption.

The same modeled stimulus produced a +0.3987% national GDP impact under normal conditions — but -0.8496% when combined with war disruption.

That difference exposed the real decision problem. The client needed to understand how the answer changed by scenario, which sectors carried the impact, and how income and spending effects reached different parts of society.

How M & Sullivan Made the Economy Testable

The engagement was designed as an analytical build-and-test process rather than a one-off calculation. Several components had to work together before the model could be trusted for policy comparison:

1.      Reconstructing a 2023 economic baseline — the latest complete 180-sector Input-Output structure from 2021 was calibrated with 2022 tax data and 2023 macroeconomic and supplementary data, including National Accounts, household surveys, investment, savings, government expenditure, transfers, FDI, fiscal data, and economic indicators.

2.      Connecting production to income and spending — sector activity was linked to labour and capital income, households, enterprises, government, capital formation, and the foreign sector, allowing the model to follow an impact beyond the production chain.

3.      Making distribution visible — Household Socio-Economic Survey data was used to divide households into five income quintiles, so policy effects could be examined across different income groups rather than disappearing inside a single national average.

4.      Balancing the system — Cross-Entropy Balancing reconciled the different datasets so every account's receipts and expenditures were consistent, while preserving official macroeconomic control totals such as GDP.

5.      Validating before simulation — the model was tested against published results for the Half-Half (Khon La Khrueng) Phase 4 measure. The close match with government reference data provided a grounded basis for applying it to new policy and shock scenarios.

Once validated, the model was used to compare four conditions: war disruption, a normal 60/40 stimulus, war combined with the stimulus, and recovery combined with the stimulus. This turned the engagement from a static impact estimate into a controlled way to ask “what changes when the context changes?”

The policy was not simply “effective” or “ineffective.” Its effect depended on the economic environment, the sectors through which it travelled, and the income and spending responses that followed.

Key Values Delivered

The engagement gave the client a structured foundation for comparing policy options under consistent assumptions. Each scenario could be examined at both industry and macroeconomic level, with the path from intervention to outcome remaining visible.

·         A balanced economy-wide baseline, integrating official data that had previously been separated by agency, definition, and release year.

·         Distributional impact analysis, showing how changes in income and consumption could affect five household income groups differently.

·         Comparable scenario outputs, covering production, value added, household income, employment, and GDP under alternative assumptions.

·         A validated analytical foundation, tested against the Half-Half Phase 4 measure before being applied to new economic shocks and policy combinations.

·         A stronger policy decision process, replacing a single headline estimate with evidence on sensitivity, distribution, sector exposure, and resilience.

Why It Worked

The SAM model worked because it forced every result to remain economically connected. A gain in one account had to appear as income or expenditure somewhere else, and macro totals had to reconcile with official controls. That is the actual value of SAM: a clearer view of how policy effects travel and how quickly the answer can change when conditions do.

Is This You?

·         You are evaluating a stimulus, industrial, tax, energy, or recovery measure with economy-wide consequences

·         Your data comes from multiple agencies or release years and does not yet form one balanced decision model

If any of that sounds familiar, that is exactly the conversation worth having before resources move.