India’s GDP revision puts factory output and input costs on a better footing

The biggest gain from India’s new GDP series is a better way to tell whether a factory produced more or merely earned a wider margin. On 27 February the National Statistical Office shifted the base year from 2011–12 to 2022–23 and released a second advance estimate of 7.6% real growth for 2025–26. The previous estimate was 7.4% on the old series. That difference combines a new method and more information. It says little by itself about which parts of the economy have changed. The more consequential revision is how the statistics convert current-price receipts and costs into real value added, and how much direct evidence they now have about enterprises that are outside the listed corporate sector. NSO’s February GDP release · MoSPI’s methods and comparison tables · Reuters, 27 February

Separating output prices from the prices of inputs

Value added is the value of gross output less the value of intermediate inputs used to make it. A factory’s selling prices can move differently from what it pays for energy, components and materials. When the price of steel, imported chemicals or fuel changes, applying one price index to nominal value added can mistake part of a price movement for a change in production. In the new GDP series, the NSO uses double deflation in manufacturing and agriculture industries where specified, and volume or single extrapolation elsewhere. Double deflation converts output and intermediate consumption to constant prices separately before taking the difference. Its accuracy depends on the quality of both price series. NSO’s February GDP release · MoSPI’s methods and comparison tables

Consider a factory with output worth ₹100 and intermediate inputs costing ₹60 in the base period. Its value added is ₹40. In the next period, suppose it sells the same quantity and uses the same volume of inputs. Its selling prices rise by 5% and the prices it pays for inputs fall by 10%. Nominal output becomes ₹105 and nominal inputs ₹54. Dividing the resulting nominal value added of ₹51 by the output-price index of 1.05 would imply real value added of about ₹48.57, more than 21% above the original ₹40. Deflating output and inputs separately gives ₹100 and ₹60, leaving real value added at ₹40. The figures are illustrative. The example shows how the change can matter particularly in industries where intermediate purchases are large relative to what the firm adds. MoSPI’s methods and comparison tables

There is an economic gain in the example. The plant earns a wider nominal margin and can spend or invest more if its costs stay lower. The change in its purchasing power and profitability is a different object from its volume of production. A firm may use the margin to hire or install equipment later, but the value-added estimate should not assign that output before it happens. Conversely, an energy shock can compress nominal margins even when physical production rises. For fiscal and monetary analysis, look at the separate effects on real activity, selling prices, profits and tax receipts.

Better evidence on small firms makes the aggregate more informative

The revised series draws on the Annual Survey of Unincorporated Sector Enterprises and the Periodic Labour Force Survey where applicable. It also incorporates more granular GST and other administrative data. This helps because the relationship between formal firms and informal firms can change. A listed manufacturer may benefit from scale, finance and access to organised buyers when smaller suppliers are struggling. Using the large firms as a proxy can miss that divergence. Direct surveys of the unincorporated sector can improve the estimate of its level and growth. The remaining questions are survey coverage, lags, and how the statistical office treats changes in reporting or shifts between organisational forms. NSO’s February GDP release · MoSPI’s methods and comparison tables

GST records are useful for a different reason. They can give a closer view of reported transactions in the registered economy. An increase in reporting can reflect higher output, an increase in prices, a change in compliance or a business moving into the tax net. Formalisation may improve working conditions or allow a firm to borrow, and those gains are worth measuring. They do not all appear as newly created physical production in the quarter when a firm starts filing. To separate the channels, compare administrative turnover with survey counts, employment and volume indicators in the same industry. The revised series is strongest when those sources cross-check one another. MoSPI’s methods and comparison tables

A higher growth rate can coexist with a smaller current-price economy

The new release puts nominal GDP in 2025–26 at ₹345.47 lakh crore and in 2024–25 at ₹318.07 lakh crore, an increase of 8.6%. The January first advance estimate on the old base put current-year GDP at ₹357.14 lakh crore and the prior year at ₹330.68 lakh crore. The new level for the current year is about ₹11.67 lakh crore, or 3.3%, smaller than the old estimate, even though the growth rate is higher. A change in the base period for constant prices cannot by itself account for a difference in current-price estimates. Updated data and methods also change the measured level. The gap is important for ratios whose numerators, such as public debt or spending, are measured separately in rupees. NSO’s February GDP release · NSO’s January first advance estimate

The figures also answer different timing questions. The new series puts real GDP growth at 7.8% in October–December, after 8.4% in July–September on the same series. Reuters described this as a deceleration. That is compatible with faster growth in October–December than in the same quarter a year earlier, as reported by the ministry. The baseline for one comparison is the previous quarter’s year-on-year growth. The baseline for the other is the year-on-year rate in the corresponding quarter of an earlier year. Analysts should specify the comparison instead of using the change in base year as evidence that the economy either accelerated or slowed. NSO’s February GDP release · Reuters, 27 February

Publish a bridge that researchers can reproduce

The strongest argument for confidence in the revision is that more direct surveys and separate price treatment address identifiable measurement problems. The strongest reason for caution is that several changes happened at once. Better inputs do not tell users how much of a given revision came from a deflator, a new enterprise estimate or late data. Parallel tables for overlapping years, sector-level contributions and eventually a documented back series would let independent researchers isolate the changes. The NSO has provided comparisons between vintages for recent years. The evidence that would build further confidence is that the revised sector estimates track independent physical output and survey indicators more consistently across expansions and downturns. That is a better test of the statistical change than the direction of the first headline revision. NSO’s February GDP release · MoSPI’s methods and comparison tables

Changes in the composition of growth are worth more attention than the two-tenths change in the annual estimate. A revised estimate of manufacturing value added can change which sector appears to be driving the expansion. Its implications for jobs depend on which industries account for the change and whether they are adding shifts, employing more workers or producing more per worker. A sector that now appears larger in measured GDP can still face weak hiring. For the same reason, the real investment series gives more policy information when read alongside capacity utilisation, employment by industry and the balance sheets of smaller firms. NSO’s February GDP release · MoSPI’s methods and comparison tables

The near-term implication is to keep the old and new estimates as distinct vintages in any model. Re-estimate fiscal ratios using a consistent nominal denominator and check whether a relationship estimated on old sector volumes still holds when new data become available. For economic policy, the questions are whether real production is growing, whether firms are investing and whether households see a gain in employment and purchasing power. Better national accounts can make those questions easier to answer. They remain separate questions.