印度20年GDP估计错误：新证据（英）2026

Abhishek Anand, Josh Felman, and Arvind Subramanian.March 2026. ABSTRACT. This working paper presents new evidence suggesting that India misestimatedits annual growth rate during the past two decades. Growth during the boomyears between 2005 and 2011 may have been underestimated by about 1–1½percentage points on average; and subsequent growth between 2012 and 2023may have been overestimated by about 1½-2 percentage points. Once theseadjustments are made, it appears that the Indian economy did not grow ata stable rate over the past two decades but rather boomed during the early2000s, then decelerated after the global financial crisis and subsequent domesticshocks. The misestimation problem can be traced to two methodological issues.The first is that the formal sector has been used as a proxy for the vast informalsector, even though unorganized enterprises were disproportionately hit after2015 by demonetization, the introduction of the goods and services tax, andthe COVID-19 pandemic. The second is that deflators for many sectors havebeen based on commodity prices, which have moved sharply in relative terms.The methodological revisions in February 2026, made following commendableconsultations, aim to address the challenges identified. Abhishek Anandis aVisiting Fellow at theMadras Institute ofDevelopment Studies. Josh Felmanis a principalat JH Consulting. Arvind Subramanianisa senior fellow at thePeterson Institute forInternational Economics. JEL Codes:E01, O47, C82, O11, O53.Keywords:GDP mismeasurement; national accounts; informal sector; pricedeflators; India. Authors’ note:Versions of this paper were presented at Cornell University, HarvardUniversity, the Indira Gandhi Institute of Development Research, the Peterson Institutefor International Economics, and the International Monetary Fund. We are grateful toparticipants at these seminars, as well as to Sajjid Chinoy, R. Nagraj, Dev Patel, JustinSandefur, and Rajeswari Sengupta for their comments. We thank Nell Henderson forexcellent editorial suggestions and Greg Auclair for checking every data point, source,table, chart, and regression. Any errors are our own. Unless we understand how the numbers are put together, and what they mean, we run the risk ofseeing problems where there are none, of missing urgent and addressable needs, of being outragedby fantasies while overlooking real horrors, and of recommending policies that are fundamentallymisconceived. — Angus Deaton, The Great Escape At the end of February 2026, following an extensive consultative process, the Indian government introduceda revised GDP.1One purpose was to update the weights of the various goods and services, a step that was longoverdue since India’s economy has gone through tremendous changes since the weights were last established in2011–12. Another purpose—perhaps an even more important one—was to address methodological shortcomingsthat had been identified by academics and statisticians. For much of the past decade, experts had raised questions about the methodology used to estimate GDP.2These doubts spread to a wider public in 2016, when the demonetization and withdrawal of 86 percent of thecountry’s currency apparently caused real GDP growth to accelerate to an eye‐popping annual rate of 8.3 percent.They resurfaced in 2019, when a credit crunch caused by a crisis in India’s nonbank financial institutions apparentlycaused only a minor blip in growth. Then, in June 2025, with private investment and job creation weak but GDPapparently booming, two former officials from the statistical agency expressed concern that “something does notadd up” (Sharma, Hussain, and Kumar 2025). Around the same time, the International Monetary Fund gave India’sGDP methodology a mere C grade. This paper revisits these debates, in order to: (a) reassess the trajectory of the world’s fifth‐largest economyduring a critical period in its development and (b) provide a benchmark for evaluating the 2026 changes to GDPmethodology. It does not assess the quality of the new numbers, something that can be done only with time. Butit does provide re‐estimated historical growth rates that could be used as a reference when the new methodologyis used to produce a GDP back series. One might ask why GDP methodology is important. After all, it hardly matters whether growth is, say, 7.2percent or 7.4 percent. If, however, the misestimation is large enough to convey a false sense of how well theeconomy is doing, serious problems arise. If the GDP numbers suggest that growth is strong when it is actuallyweak, businesses are liable to misinvest, households to overspend, and the central bank to maintain an excessivelytight monetary policy. Inaccurate numbers also make it difficult for the government to calibrate its fiscal or reformpolicies, as it cannot respond to problems it cannot see. For all these reasons, getting GDP growth right is criticallyimportant. This paper examines the GDP methodology introduced in Janua