Yesterday, the central bank released its Annual State of the Economy report, and the state of the agricultural GDP has little to write home about. The sectoral performance was covered in some detail by BR Research back in May and June when provisional GDP estimates were released by National Accounts Committee, and MoF through Annual Economic Survey, respectively. (For more, read ‘Agri-GDP if wishes were livestock’, published on 20 May 2020; and, ‘Agri-GDP: a house of cards’, published on 12 June 2020).
Unlike other real sectors of the economy where the estimates released by SBP are revised to take into account the final tally for the last fiscal quarter – such as the monthly Large Scale Manufacturing Index which is released at 45 day intervals – estimates for the agricultural sector are rarely revised between the two publications. The intuitive explanation for the practice is that agriculture output estimates are already finalized by the time Annual Economic Survey is released in June, and as such do not require revision later when the central bank releases its annual publication on economic performance.
Except, agricultural outputs are revised, especially those of the cropping segment. Every subsequent Economic Survey publication uses revised estimates for crop production in the preceding year, as does the SBP’s annual State of the Economy. Afterall, those estimates are reported as ‘provisional’ with good reason. Which raises a question: why are these estimates not revised during the long 100 - 150 days interval between the publication of Economic Survey and State of the Economy reports?
Turns out, both publications list Pakistan Bureau of Statistics as the source for estimates of agricultural dataset. Recall that the original provisional estimates are also announced by PBS in May through the National Accounts Committee, indicating that PBS either does not receive, compile, or report the revised data at least until November of the following fiscal year.
A lag in revision of data would be fine and in line with delays that are characteristic of Pakistan’s bureaucratic speed; except, the annual timeline of agricultural data collection insists that there may be more to the story. Consider that most of the major crop output is already available latest by August, when kharif crops (rice, cotton, cane, and maize) of the preceding season have already been processed and milled, whereas winter (wheat) and spring (maize) crops have also been harvested. This implies that any revisions in provisional estimates of output levels should be available for reporting by November.
Instead, could the year long wait in revision appears more to be a ‘desired’ adjustment of output level in the base period? Consider that extent of annual revision is low for rice and cotton. While no robust mechanism for independent rice harvest and milling exists other than that reported by provincial crop reporting services, the country produces as much as 50 percent exportable surplus in rice, and as such has ensured that crop output level remains insulated by political need for revision.
On the other hand, cotton output level has also been rarely revised, despite its falling production. This may be attributable to the relatively independent data collection system in place in the form Cotton Crop Assessment Committee, where private sector members come together with provincial and federal representatives to report a consensus estimate.
Compare this to crops where base period output level has been revised significantly in recent years. The Sugar Inquiry Committee has already brought to light that no independent mechanism exists to measure sugarcane crop production data; instead, the country relies on crushing estimates as reported by the Pakistan Sugar Milling Association (PSMA). For wheat, data varies significantly, even as government is the biggest buyer in the market, and has a heavy footprint in crops monitoring, movement (by way of release to mills), and transport.
Pakistan’s agricultural GDP estimation suffers from a serious trust deficit problem in urgent need of redressal. Already, little evidence-based research is conducted to estimate nearly two-thirds of remainder agricultural GDP which includes livestock, fishing, and forestry segments – and whose annual output is extrapolated based on census conducted 10 and 15 years ago. Routine media reports that the government exaggerates output estimates for crops such as cotton and wheat also don’t help. With every economic publication referencing the apex statistical body as the data source, the folks at PBS should take note.