Taiwan Macroeconomic Model (BVAR) — Excel-Based Forecasting & Scenario Analysis

A Windows desktop forecasting tool built on a Bayesian VAR (BVAR) model for Taiwan’s key macroeconomic indicators.
It is designed for non-technical researchers to conduct forecasting and scenario/conditional analysis via Excel-based conditioning—no coding and no R installation required.

---

Key Features

• BVAR model for Taiwan macro forecasting
• Unconditional (baseline) forecast generation
• Conditional/scenario forecasts by editing an Excel sheet (forecast_condition)
• Exports results to:
  • result_raw.xlsx (baseline + conditioning + conditional forecast)
  • result_summary.xlsx (level, growth, annual summaries)
• Users can re-run Part 2 repeatedly after adjusting scenarios in Excel

---

Installation (Windows)

1. Go to the Releases page and download the latest installer:
   • https://github.com/TonyLoWaiYau/Taiwan-Macroeconomic-Model/releases/latest
2. Run the installer and launch the application from the Start Menu (or desktop shortcut if created).

Notes:

• Microsoft Excel (or compatible .xlsx editor) is required for scenario editing.
• The installer bundles everything needed to run (users do not need R or RStudio).

---

How to Use (Workflow)

Part 1 — Baseline forecast + create conditioning template

1. Open the app.
2. Upload your data.xlsx (see required format below).
3. Click Run Part 1 (Create result_raw.xlsx).
4. Click Open result_raw.xlsx in Excel.

This creates result_raw.xlsx containing:

• result_raw (history + baseline forecast)
• forecast_condition (editable sheet for conditional/scenario paths)

Part 2 — Conditional/scenario forecast (repeatable)

1. In Excel, edit ONLY the forecast_condition sheet (future rows).
2. Save and close Excel (the file must not be locked).
3. In the app, click Run Part 2 (Conditional Forecast + result_summary.xlsx).

If the scenario is problematic (e.g., non-numeric values), the app will show an error message.
You can then edit forecast_condition again and re-run Part 2 until satisfied.

---

Input Data Format: data.xlsx

• The workbook must contain a sheet named clean (default; configurable in the app).
• Data must be quarterly and include the following variables.

Variables

Variable	Description
t	Time (quarterly)
gdp	Real GDP in million TWD (2021 prices)
pce	Real Private Consumption Expenditure in million TWD (2021 prices)
gce	Real Government Consumption Expenditure in million TWD (2021 prices)
gcf	Real Gross Capital Formation in million TWD (2021 prices)
exports	Real Exports in million TWD (2021 prices)
imports	Real Imports in million TWD (2021 prices)
g_gdp	Advance estimate of real GDP growth (% YoY)
g_pce	Advance estimate of real PCE growth (% YoY)
g_gce	Advance estimate of real GCE growth (% YoY)
g_gcf	Advance estimate of real GCF growth (% YoY)
g_exports	Advance estimate of real Exports growth (% YoY)
g_imports	Advance estimate of real Imports growth (% YoY)
ip	Industrial production index (2021 = 100, quarterly average)
cpi	CPI (2021 = 100, quarterly average)
unemploy	Unemployment rate (quarterly average)
fx	USD/NTD (quarterly average)

Notes on “advance estimate” variables (g_*)

• The model computes YoY growth internally from level variables.
• The g_* columns are used to optionally provide advance estimates for the latest quarter (if the last row of the input data has missing values).

---

Output Files

result_raw.xlsx

Contains:

• result_raw: historical data + baseline forecast
• forecast_condition: scenario input sheet (editable)
• conditional forecast: written/updated after running Part 2

result_summary.xlsx

Contains:

• result_level: quarterly levels (including forecast horizon)
• result_growth: quarterly YoY growth rates (levels converted to growth)
• result_annual_level: annual aggregates (sum or average depending on variable)
• result_annual_growth: annual YoY growth rates

---

Troubleshooting

• Part 2 fails / shows an error:
  Usually caused by non-numeric cells in the future rows of forecast_condition, or by unrealistic conditioning paths that break the conditional forecast solver.
  Fix the sheet, save and close Excel, then try Part 2 again.

• Cannot overwrite result_raw.xlsx:
  Ensure Excel is closed (file not locked).

---

Author

Tony Lo (Feb 2026)