India\u2019s telecom firms have been badly hit by government policy. Initially, spectrum was allocated free, but in return for a revenue-share-based licence fee; when that was replaced by auctions for spectrum, the licence fee was not abolished. As a result, the share of government revenue by way of recurring licence\/spectrum\/auction fees rose from 11% in FY07 to 32.4% in FY17. This was compounded by not auctioning enough spectrum and, at critical times, telling telcos their licences would be renewed only if they were able to buy fresh spectrum. Amazingly, the valuations bid for in these constrained-auctions were, by and large, used as the reserve prices in the next round of auction. So, in 2010, when 3G spectrum was auctioned, limited spectrum and industry exuberance ensured that in Delhi, for instance, while the reserve price for a 5 MHz slot in the 2100 MHz frequency band was Rs 64 crore per MHz, the winning bid was for Rs 663 crore. And when, in 2012, auctions were carried out for 1800MHz spectrum that telcos were planning to use for primarily voice traffic\u20143G spectrum was primarily for data services\u2014the reserve price was at 3G-type levels of Rs 554 crore per MHz. As a result, India\u2019s spectrum prices are much higher than those globally. In the 600MHz band which was auctioned in the US last year, the auction value was Rs 3,480 crore per MHz. In India, by contrast, the 700 MHz band has a reserve price of Rs 11,485 crore per MHz. Add in the fact that every customer in the US has a monthly telecom bill that is many times more than in India, and the equation becomes really unfavourable. Hardly surprising, then, that there have been no bids for the 700 MHz band. Not surprisingly, while this policy hit telcos, it also hit the government. In 2012, this resulted in 67% of the spectrum on auction not getting sold, and this rose to 80% in 2013 and 100% in 2015; it was 59% in 2016 and an average of 38% since the 2010 auction. In FY17, the government budgeted Rs 98,994 crore from telecom revenues including auctions but got Rs 78,715 crore; with the industry almost bankrupted (even before RJio came in), it held no auction in FY18 and budgeted just Rs 44,342 crore from the sector, and there is likely to be a shortfall of Rs 15,000 crore in even that. So how do you calculate the right reserve price, Trai asked in an open house last week? This is critical since there have been, over the years, several allegations of favouritism with the way Trai\/DoT fix the reserve price. And the current way of using the last auction value to, by and large, set the reserve price means the distortions caused by either spectrum shortages or irrational exuberance get carried over to the next auction. What Trai\/DoT need to do is to work on a scientific way to calculate the price, to replace the arbitrariness that has been the norm so far\u2014when reserve prices are cut after a failed auction, taking a call on how much the cut should be is also arbitrary. Broadband India Forum (BIF) has worked on a rough model to determine reserve prices in a more rational manner. To begin with, the reserve price must take into account relative propagation qualities of each band. BIF takes the 800 MHz band as the reference model, but it makes no difference if some other band, 1800 say, is used as the reference. Its model assumes 700 MHz is 15% more efficient and 1800 MHz just 48% as efficient; other bands have been compared in a similar fashion. Since all valuations have to be done at today\u2019s prices, BIF takes the historical auction price in each band and each circle and, based on inflation, inflates them to arrive at today\u2019s price. The relative propagation characteristics are also baked in since bands with better propagation should be priced higher. Since it is important to remove distortions of past auctions, BIF\u2019s model also factors in the revenue potential of the spectrum in each circle. The fact that, in 2010, the Delhi and Mumbai circles accounted for just 14% of telco revenues but got 40% of the auction values bid is surely problematic. It is possible that, in the auction, telcos may once again bid exorbitantly for these circles, but it is the job of Trai to correct for this each time and then leave it to the market which takes into account demand for a service, the availability of devices in each band, etc. Having arrived at a notional value of the spectrum, BIF assumes the reserve price should be 50% of this, but DoT\/Trai can just as well work on 60%, 70%, whatever\u2014ideally, the lower the reserve price, the more scope the market has for genuine price discovery. Based on that (see graphic), the reserve price of the 700 MHz band need to be cut by 65% of what it is today, 800 MHz by 43%, 900 MHz by 63%, 1800 MHz by 51%, 2100 MHz by 63%, 2300 MHz by 27% and 2500 MHz by 31%. The beauty of this method is that even if telcos are irrational in the next bid or there is a spectrum shortage, the price distortion can be corrected in the next auction. Once Trai adopts a model like this, there will be no allegations of favouritism in the future and no CAG-type investigations of notional losses caused to the exchequer by reserve prices being either too low or too high.