Quick answer
Economic Survey 2025-26 nowcasts Q3 FY26 real GDP growth at 7%, using high-frequency indicators, a Dynamic Factor Model and ARIMAX.
Nowcasting estimates current or very recent economic activity before official statistics arrive.
Box I.2 reports a Q3 FY26 real GDP growth nowcast of 7%, using information through December 2025; it is not an official quarterly estimate.
Concept diagram
Reading the economy before GDP arrives
A radar gathers frequent economic signals for a timely estimate of current activity.
- Nowcasting estimates the present before official GDP is available.
- Model estimates remain uncertain and can change with incoming data.
Basics: signals, factors and prediction
Information advantage
- Quarterly GDP is released two months after the quarter ends; monthly indicators provide earlier signals.
- The dataset begins in January 2006.
Latent common factor
- A Dynamic Factor Model separates shared movement across indicators from series-specific noise.
- A scree plot helps choose a parsimonious number of factors.
ARIMAX
- An autoregressive model uses past behaviour; exogenous factors add information from outside the target series.
- The Survey uses AR(1), integration order 0 and MA(1), with a dummy for extreme growth periods.
Box I.2: the estimation pipeline
Prepare HFIs
The box reports 17 indicators. Non-PMI variables are transformed to YoY growth using log differences; PMIs enter as levels.
Extract common movement
Standardised indicators enter a DFM; factor dynamics are represented in state-space form.
Bridge frequencies
Average monthly factor values within each quarter, then use the quarterly factor in ARIMAX.
Data chart
Chart I.19: early factors capture most common variation
- The first factor accounts for more than half the variation.
- Later factors add progressively less information; adding factors is not automatically an improvement.
Source: Economic Survey 2025-26, Chart I.19 (author’s calculations) · Values read from the Survey chart; approximate.
Show the data (Per cent)
| Period / item | Cumulative explained variance |
|---|---|
| 1 factor | 58 |
| 4 factors | 80 |
| 7 factors | 90 |
Model validation and output: Box I.2
| Measure | Reported result | Meaning |
|---|---|---|
| In-sample RMSE | 1.6 | Fit over the estimation sample |
| Out-of-sample RMSE | 0.87 | Forecast error outside the fitted sample |
| Portmanteau test | Residuals are white noise | No remaining serial pattern detected by this diagnostic |
| Q3 FY26 nowcast | 7% | Based on data through December 2025 |
Exam traps: distinguish the concepts
Statements examiners like to twist — learn the correct version.
Nowcasting replaces official GDP measurement.
It supplies an early model-based estimate, not a replacement for national accounts.
Out-of-sample RMSE must exceed in-sample RMSE.
The reported values are 0.87 and 1.6 respectively; sample composition matters.
All HFIs are converted to growth rates.
PMI manufacturing and PMI services enter as index levels.
Prelims focus
- DFM → monthly factors → quarterly averages → ARIMAX.
- A scree plot concerns explained variance, not the GDP growth rate.
Mains angle
- Evaluate the usefulness and limitations of high-frequency nowcasts in policymaking.
- Discuss data revisions, structural breaks and model uncertainty before interpreting a point estimate.
Flip & recall
Answer in your head first, then tap the card to check.
Practice MCQs
Original Prelims-style questions with explanations.
0 / 3 correct · 3 left
Q1.Nowcasting mainly estimates:
Q2.Monthly factors are converted to quarterly factors by:
Q3.The reported out-of-sample RMSE is:
