Why This Part Is the Test
This is an advocacy document. It is the case for the prosecution against Congress governments in power. The selection of material is one-sided on purpose. The evidence in it is not. Every serious claim carries a grade showing how well it is proved, the other side is given its strongest argument, and where the other side wins, this book says so. A second series applying the same test to the government of 2014 onwards is planned.
Part One of this series set out a scale for grading evidence, and it used one case as its worked example. That case is in this part.
The reasoning was set out at the beginning and it has governed fifteen parts since. It is worth repeating here because this is where it earns its keep or fails.
The Comptroller and Auditor General reported in November 2010 that the allocation of 2G telecom spectrum had cost the exchequer up to one lakh seventy-six thousand crore rupees. In February 2012 the Supreme Court cancelled a hundred and twenty-two licences. In December 2017 a criminal court in Delhi acquitted every single person accused, and said the prosecution had produced no evidence of criminality at all.
All three of those things happened. A book that gives you the first two and not the third is not a prosecution document; it is a pamphlet, and the first reader who knows about the acquittal stops believing anything else in it.
Every account of these five years assumes that the size of a scandal is the number attached to it. One lakh seventy-six thousand crore. One lakh eighty-six thousand crore. The numbers are how these events are named.
But the figures are not money that went missing. Nobody removed one lakh seventy-six thousand crore rupees from a treasury. The number is a presumptive loss — an auditor’s estimate of the difference between what the government charged and what it might have raised had it done something else.
Which means the number is not a measurement of a theft. It is a measurement of a policy choice, made by comparing it with a policy that was not chosen. Chapter Three is entirely about that sentence.
This matters in both directions, and this book is going to insist on both. A presumptive loss is not proof of a crime, and treating it as one is how the criminal cases collapsed. It is also not nothing — a government that gives away a public asset below its value has done something real to the public, whether or not anybody took a bribe.
Nearly every argument about these five years, on both sides, consists of pretending one of those two halves does not exist.
So Part Sixteen is arranged differently from the rest of the series. Chapter Three explains what the numbers are before any of the cases are described, because a reader who does not have that will not be able to follow Chapter Eight, which sets out what was actually established and what was not.
What is in this part and what is not
The four subjects agreed for this part are 2G, Coalgate, the Commonwealth Games and Adarsh, together with what the courts found afterwards. They are all here.
Part Fifteen deliberately kept out the money — including the spectrum allocation of January 2008, which happened under UPA I. That was so this part could take the whole thing at once rather than telling half of it twice. The 2008 allocation is therefore in Chapter Four, where it belongs.
1The Government That Won
In May 2009 the Congress won two hundred and six seats — sixty-one more than in 2004, and its best result since 1991. The alliance took two hundred and sixty-two. For the first time since 1984 a Congress government did not need support from outside to survive a vote.
Part Fifteen ended by noting how strange that result was against the record it followed. This part begins with what was done with it.
1.1 — What the position was
Manmohan Singh returned as Prime Minister. The arrangement described in Part Fifteen was unchanged in shape: a Prime Minister without a party base, a party president outside government, and a National Advisory Council advising on legislation from outside the Cabinet.
What changed was the arithmetic. The Left parties, whose withdrawal in July 2008 had produced the confidence vote of Part Fifteen’s Chapter Eight, were reduced from around sixty seats to about twenty-four. They no longer mattered.
From 2004 to 2008 the government needed roughly sixty Communist votes to pass anything, and lost them in the end.
From 2009 it needed nobody in particular. The largest single coalition partner had around eighteen seats.
This is the strongest position a Congress government had been in for twenty-five years, and everything in this part happened from inside it. Where Part Fifteen’s defence could point to weakness, this one cannot.
1.2 — What the majority was for
A government that wins sixty-one extra seats usually arrives with a plan for them. This one did, and it is worth recording because almost none of it happened.
The reading inside the government was that the 2009 result had rewarded the welfare legislation of the first term — the employment guarantee above all — and that the second term should do two things: extend that programme of rights, and return to the economic reforms that the Left parties had blocked between 2004 and 2008.
The first half was done. Chapter Two is the list.
The second half was not. Insurance, pensions, retail, banking, labour law, subsidies, a goods and services tax — the agenda that had been held up by allies who no longer had the votes to hold anything up — mostly did not move.
The defence says: a government elected on welfare legislation did more welfare legislation. That is not a failure of nerve; it is doing what the mandate said. And the economic reform agenda is contested — several of the items on it were opposed by serious people for serious reasons, and the goods and services tax in particular required the states to agree, which took another six years and a constitutional amendment under a different government.
The reply: the goods and services tax point is fair and the state-consent problem was real. The rest is not. Insurance and pensions were held up for four years by parties with sixty seats and then not moved when those parties had twenty-four. Something other than arithmetic was operating after 2010, and Chapter Seven says what it was.
Where this matters for the whole part: Part Fourteen’s government had 232 seats and no majority and reformed the economy in thirty-three days. This one had the strongest position since 1984 and did not. Whatever explains these five years, it is not that the government lacked the votes.
Set the three Congress governments this series has covered in sequence beside one another and the point becomes hard to avoid.
Compare the three Congress governments this series has now covered in sequence.
Part Fourteen: 232 seats, no majority, three weeks of foreign money. It dismantled the licence system in thirty-three days.
Part Fifteen: 145 seats, dependent on Communist parties with a veto. It passed the transparency law and the employment guarantee.
Part Sixteen: 206 seats, no dependence on anybody. It passed the food and education laws — and then, from about 2011, largely stopped.
The correlation between parliamentary strength and what a government achieves, across these three, is either zero or negative. That is worth holding on to, because it is the opposite of what every account of Indian politics assumes.
1.3 — The two things that arrived together
Two developments outside the government’s control shaped the whole term, and both were consequences of things it had done in its first.
The first was the Right to Information Act of 2005. By 2009 it had been running for four years and people had learned to use it. A very large amount of what appears in this part — file notings, correspondence between ministries, the internal record of decisions — became public because of a law this government had passed. Part Fifteen made that point about the employment programme. It applies here on a much larger scale.
The second was the Comptroller and Auditor General, and specifically what that office did between 2010 and 2012.
The CAG is the constitutional auditor of the Indian state. It examines whether public money was spent lawfully and whether public assets were handled properly, and reports to Parliament.
It is not a court and it is not an investigating agency. It cannot charge anybody, summon anybody or find anybody guilty. Its reports go to Parliament’s Public Accounts Committee.
Part One’s grading scale has a rung called Audited precisely because this office exists: a finding that is official, serious, and not a legal determination of anything.
Between November 2010 and August 2012 the CAG produced reports on telecom spectrum, on coal, and on the Commonwealth Games, and those three documents are the origin of almost everything in the six chapters that follow.
It is worth being exact about where this part’s material comes from, because it is unusual.
Nearly all of it is the government’s own paper: audit reports laid before Parliament, ministry file notings obtained under the transparency law, Supreme Court judgments, and charge sheets. Very little of it is journalism and almost none of it is anybody’s memoir.
Part Fourteen had to build its central charge out of one court affidavit and one lawyer’s autobiography, because the file notes of 1992 have never been released. Part Sixteen has the opposite problem: too much documentation rather than too little, and Chapter Three is about how a mass of official paper can produce a false picture as easily as a shortage of it.
Those three documents, and the law that made the underlying files readable, are why this part exists in the form it does.
In 2009 the Congress won 206 seats — its best since 1991 — and for the first time since 1984 did not need anybody’s outside support.
Everything in this part happened from a position of strength. The weakness defence that Part Fourteen and Part Fifteen could both use is not available here.
And almost every document in this part is public because of a law this same government passed in 2005.
2What Was Built
Parts Fourteen and Fifteen both put the credit chapter second and said why: at the end it reads as a sweetener produced late. The same applies here, and here it is harder, because this is the part everybody knows as the scandals part.
2.1 — Two more rights
Part Fifteen described the distinction this government kept returning to: between a scheme, which stops when the money runs out, and a right, which a named person can enforce. UPA II added two more.
The Right of Children to Free and Compulsory Education Act, 2009 made schooling from six to fourteen an enforceable right, gave effect to a constitutional amendment passed in 2002, and required private schools to reserve a quarter of their places for children from disadvantaged backgrounds.
The National Food Security Act, 2013 gave a legal entitlement to subsidised grain to roughly two-thirds of the population.
The food law covered around eight hundred million people.
That is more people than live in the whole of Europe. It is the largest food entitlement programme any country has ever legislated.
Whether it worked is a separate question and the answer is mixed. But the scale is worth sitting with before reading the rest of this part, because the same government, in the same years, is the subject of Chapters Four to Six.
Both have records, and this book gives them.
Enrolment was not the education law’s problem. By the time it passed, well over ninety per cent of Indian children were already enrolled in primary school.
What the surveys of the following decade found was that a large proportion of children in Class Five could not read a Class Two text or do simple division — and that this did not improve much, in some measures got worse, and was not much affected by the Act.
The law guaranteed a place in a school. It did not guarantee that anything happened there, and the provisions most likely to have addressed that — on teacher accountability and on assessment — were the weakest in it.
The quarter of private school places reserved for disadvantaged children is real and is used. It is also the part most litigated and least uniformly implemented.
The food law’s record is better on its own terms and is harder to separate from what existed before it, because most states already ran subsidised grain distribution and the Act extended and entrenched rather than invented.
2.2 — Three more laws, and where they came from
The Land Acquisition Act, 2013 replaced the colonial statute of 1894 that Part Thirteen described in detail — the Act under which land could be taken with no requirement of consent, and which was the first of the five mechanisms Part Thirteen listed by which tribal land was lawfully taken away. The 2013 Act required consent from a proportion of affected landowners, a social impact assessment, and compensation at a multiple of market value.
That is a debt of this series paid. Part Thirteen named the 1894 Act as the instrument and noted it stayed in force until 2013. This is where it went.
The Lokpal and Lokayuktas Act, 2013 created a national anti-corruption ombudsman with jurisdiction over the Prime Minister. The Criminal Law (Amendment) Act, 2013, following the Justice Verma Committee report after the Delhi gang rape of December 2012, rewrote the law on sexual offences.
Two of those five have an awkward origin, and this book is not going to hide it.
The case against giving this government the credit: the Lokpal Act was passed in December 2013 after two years of a mass movement demanding it, hunger strikes, and a collapse in the government’s standing. The criminal law amendment was passed after nationwide protests. The food law was passed months before a general election.
On this reading these are not achievements but concessions — things extracted from a government that had run out of room, timed for a campaign.
The reply: the timing is real and the pressure is real, and the record should say so. But Part Fourteen made the same point in reverse about the 1991 reforms, which were also done under duress, and concluded that being forced to act does not tell you what a government will do when forced. Governments under pressure frequently pass nothing, or pass a version with the teeth removed.
Where the criticism does land: on the Lokpal specifically. The Act was passed in December 2013. The first Lokpal was not appointed until March 2019 — under a different government, and after the Supreme Court had been asked twice to make somebody do it. A law that takes five years to produce an office is a law that was passed and not implemented, and the government that passed it had five months left to start.
There is a way to settle that argument which does not depend on anybody’s account of their own motives.
The awkward question about all five of these laws is whether they were this government’s doing or things it was made to do. There is a way to test it.
Ask which of them a successor government tried to undo. The land acquisition law of 2013 was the subject of an ordinance in December 2014, re-promulgated twice, seeking to remove the consent requirement and the social impact assessment for several categories of project. It was abandoned in 2015 after it could not be passed.
A law that the next government spends a year trying to weaken and cannot is a law with real content in it. That is a harder test than any statement of intent, and the 2013 Act passes it.
The Lokpal Act does not pass the equivalent test, for the opposite reason. Nobody had to weaken it. It simply was not constituted for five years.
2.3 — The thing that was built without a law
The largest single thing this government created is not on the list above, because it was not a statute at all.
The Unique Identification Authority of India was set up by executive order in January 2009, under Nandan Nilekani, a technology executive brought in from outside government. The first Aadhaar number was issued in September 2010. By the time this government left office in 2014, something in the region of six hundred million people had been enrolled.
In January 2013 the government began direct benefit transfers — paying subsidies and pensions into bank accounts rather than distributing them through intermediaries.
Six hundred million people in about four years. That is roughly four hundred thousand enrolments a day, every day, including fingerprints and iris scans, in a country where a large proportion of those enrolled had never held any identity document at all.
It is the largest biometric identity programme ever attempted anywhere, and it was built by an authority that had no statute behind it for seven years.
The Aadhaar Act was passed in 2016, under a different government, and the Supreme Court ruled on its constitutionality in 2018 — by which time it was already the basis on which most Indians received most things from the state.
Both the scale and the way it was achieved drew objections at the time, and the two are not separable.
The objection, which is serious and was made at the time by people on both sides of politics: a body collecting the biometric data of most of the population, created by executive order, operating for seven years without a law defining what it could do with that data or who could demand it, is exactly the kind of thing a Parliament exists to authorise before it happens rather than afterwards.
Chapter Seven describes a Parliament that lost thirty per cent of its sitting time. This is what a government does when it wants something large and cannot get a bill through: it does it without one.
The defence: a bill was introduced in 2010 and rejected by a parliamentary standing committee in 2011. The programme was voluntary in law throughout this period. And the benefit — a poor person being able to prove who she is, and receive money without an intermediary taking a share — is very large and was arriving while the argument continued.
Where this leaves it: the achievement is real and so is the objection, and they are the same fact. The speed came from not needing anybody’s permission, and that is also the criticism.
2.4 — What was not built
The communal violence law promised in the Common Minimum Programme of 2004, drafted in 2005 and again in 2011, was still not enacted when this government left office in May 2014. Part Fifteen’s Chapter Nine set out what it would have done. Ten years, two terms and two comfortable majorities.
Public spending on health remained close to one per cent of national income across both terms, against a promise of two to three.
Substantial, and mostly real. Education and food as enforceable rights; the 1894 land acquisition Act finally replaced; a rewritten law on sexual offences. Together with Part Fifteen’s transparency and employment laws, this is the largest body of rights-creating legislation by any Indian government since the 1950s.
The qualifications, all of which stand: several were passed under heavy public pressure and close to an election; the Lokpal was legislated and then not constituted; and the two commitments requiring sustained money rather than a statute — health and education spending — were not met in ten years.
Which is a stronger record than the reputation of these five years suggests, and a weaker one than the list of statutes on its own implies.
UPA II made education a right, made subsidised food a right for around eight hundred million people, replaced the 1894 land acquisition law that Part Thirteen named as the instrument, and rewrote the law on sexual offences.
Several of those came under enormous public pressure and close to an election. The Lokpal Act was passed in December 2013 and no Lokpal was appointed until March 2019.
3The Number
This chapter contains no scandal. It explains a method of calculation, and everything in the five chapters that follow depends on it.
3.1 — What a presumptive loss is
Suppose a government owns something valuable — telecom spectrum, a coal seam, a piece of land — and gives it to a company for a fixed fee.
An auditor asks: what would this have fetched if it had been auctioned? He finds a comparable auction, works out a price, multiplies it by the quantity given away, subtracts what was actually charged, and reports the difference.
The gap between what the government received and what an auditor calculates it could have received under a different method of allocation.
It is called presumptive because the alternative did not happen. The auditor is comparing a real transaction with an imagined one.
Three things follow, and they are the whole chapter. It is a real and legitimate audit technique. The figure depends entirely on which comparison you choose. And it says nothing whatever about whether anybody was paid a bribe.
3.2 — Why the figures move so much
If the number depends on the comparison chosen, then choosing a different comparison gives a different number. That is not a scandal about auditors; it is arithmetic. But the size of the movement in these two cases is worth seeing set out.
| Case | Figure | Who produced it, and on what comparison |
|---|---|---|
| Spectrum | ₹1,76,645 crore | CAG, November 2010 — the highest of several estimates it gave, based on the 3G auction price of 2010 |
| Spectrum | ₹57,666 crore | CAG, same report, on a different comparison — an offer made by one company for a stake |
| Spectrum | ₹30,984 crore | The CBI’s own charge sheet. The prosecuting agency’s figure, used in court |
| Coal | ₹10.67 lakh crore | CAG draft report, March 2012, which leaked before publication |
| Coal | ₹1.86 lakh crore | CAG final report tabled in Parliament, August 2012 — the draft figure divided by about five and a half |
The coal draft leaked in March 2012 and the final report came in August. For five months the country argued about a figure the auditor did not ultimately publish. The two numbers differ by roughly nine lakh crore rupees, which is more than the entire central government budget of that year.
Look at the third row. It is the most useful fact in this part and it is almost never quoted.
The Central Bureau of Investigation, prosecuting the spectrum case, put the loss at ₹30,984 crore.
That is about seventeen and a half per cent of the audit figure the whole country was arguing about.
The agency trying to convict people used a number roughly one-sixth the size of the one used to describe the scandal. It did that because a charge sheet has to survive cross-examination and an audit report does not.
None of which is a complaint about the office that produced the figures.
It would be easy to read this chapter as a criticism of the auditor. It is not, and the distinction matters.
The CAG’s spectrum report did not give one figure. It gave several, on different comparisons, and said what each rested on — including the ₹57,666 crore figure in the second row of the table above. The qualification that the loss was presumptive was in the report.
What happened to those numbers happened outside the audit. The highest figure was extracted, quoted without its basis, and became the name of the affair. The auditor of the day later wrote a book defending the estimate, which is his right, and the office’s reputation was drawn into a political fight that the report itself had not started.
A reader who takes from this chapter that the CAG behaved improperly has taken the wrong thing. The failure was in the reading, and this book was among the many that did the reading badly for years.
3.3 — What it can and cannot prove
Here is the distinction that the next five chapters turn on, and it is not a technicality.
| The question | Who answers it | Standard of proof | What follows |
|---|---|---|---|
| Was this allocation lawful and proper? | Auditor, then a constitutional court | Was the process arbitrary or unreasonable? | The allocation can be cancelled |
| Did a named person commit a crime? | A criminal court | Beyond reasonable doubt, against that individual | That person goes to prison |
These are the two questions that were repeatedly confused between 2010 and 2018, by newspapers, by politicians and — Chapter Eight argues — by the investigating agencies themselves.
A government can give away a public asset far below its value, entirely without corruption, through ordinary bad judgement or an ideological preference for cheap spectrum over expensive spectrum. The first question would be answered against it. The second would not arise.
Equally, an allocation can be perfectly defensible in policy terms and still involve somebody taking money to arrange it. The second question would be answered against him and the first would not.
The two answers are independent. That is why the Supreme Court could cancel a hundred and twenty-two licences in 2012 and a criminal court could acquit everybody in 2017 without either being wrong.
The objection to this whole chapter: that it is a sophisticated way of letting a government off. Everybody knew what happened. Companies with no telecom experience obtained licences at 2001 prices in 2008 and sold stakes within months at enormous multiples. Whatever a criminal court later said about the evidence, the public understood the transaction correctly, and a book that spends a chapter on audit methodology is helping to bury it.
This objection has real force and the reply is not comfortable. The stake sales happened; Chapter Four gives the figures. The public reaction was not irrational.
But the objection proves the chapter’s point rather than defeating it. “Everybody knew” is exactly the standard this series has spent fifteen parts refusing. Part Eleven declined to name individuals against whom no body had found. Part Twelve refused to assert that Rajiv Gandhi ordered the Ayodhya locks opened or took Bofors money. Part Fourteen refused to say the central government wanted the demolition.
A book that applied those rules for fifteen parts and abandoned them here — at the one place where abandoning them would be popular — would be worth nothing.
3.4 — What survives
Strip out what the presumptive-loss figure cannot support and something is still left standing, and it is not small.
Audited, and in the two largest cases subsequently Court-found:
That public assets of very large value were allocated by discretionary methods rather than by auction. Not disputed by anybody, including the government at the time.
That the methods used were arbitrary in their operation. Found by the Supreme Court in both the spectrum case (2012) and the coal case (2014).
That the beneficiaries obtained assets worth substantially more than they paid. Demonstrated by what several of them did next.
What the figures do not establish is that anybody was bribed. That is a different question, it was tried separately, and Chapter Eight gives the results.
Everything from here on should be read against that list, because the five chapters that follow are an attempt to say which side of it each case falls on.
A presumptive loss is the gap between what a government charged and what an auditor calculates it could have charged under a method it did not use. It is a real technique. It is not a measurement of theft.
The spectrum figure ran from ₹1.76 lakh crore in the audit to ₹30,984 crore in the prosecution’s own charge sheet. The coal figure fell from ₹10.67 lakh crore in the leaked draft to ₹1.86 lakh crore in the published report.
Two questions, always. Was the allocation proper? Did somebody commit a crime? They have different answers and different judges, and confusing them is what wrecked the next five years.
4Spectrum
On the tenth of January 2008, the Department of Telecommunications issued a hundred and twenty-two new telecom licences with spectrum attached, at a price fixed in 2001, on a first-come-first-served basis. The Minister for Communications was A. Raja of the Dravida Munnetra Kazhagam, a coalition partner.
This chapter sets out what was done, what was found, and what happened to all of it.
4.1 — What spectrum is and why the price matters
Radio frequencies. A mobile phone call travels on one, and only a finite range exists. It cannot be manufactured, and every operator needs some.
It belongs to the state. Governments either auction it to the highest bidder or hand it out at an administered price on conditions.
The argument for auctioning is that it captures the value for the public. The argument against is that operators who pay enormous sums pass the cost to subscribers, which in a poor country slows the spread of telephones. Both arguments are respectable, and the second one was Indian policy for most of the previous decade.
4.2 — What was done
The mechanics matter, because this is where the case was later argued.
A cut-off date for applications was announced as the first of October 2007. Applications received after that date were not to be considered. In January 2008 the cut-off was changed to the twenty-fifth of September, which excluded a number of applicants who had applied in between.
Companies were then told, at short notice on the tenth of January, to bring demand drafts to the department. Those who arrived first got licences. The price was the one fixed in 2001, when there were around four million mobile subscribers in India.
India had about four million mobile subscribers when that price was set in 2001.
By January 2008 it had around two hundred and thirty million, and was adding roughly eight million a month.
The asset had become somewhere between fifty and a hundred times more useful, and the price had not moved at all. That is the fact from which everything else in this chapter follows, and it is not in dispute between any of the parties.
Two of the companies that obtained licences had no telecom operations. Within months, both sold substantial stakes to foreign telecom groups.
The stake sales are the most solid evidence in this chapter because they are commercial transactions between unconnected parties, filed with regulators, and nobody has ever disputed them.
One company had paid about ₹1,537 crore for its licences and sold around forty-five per cent of itself to a Norwegian operator months later, in a transaction valuing it far above that. Another, which paid a comparable amount, sold a majority stake to a company from the Gulf.
An arm’s-length buyer paying a large multiple within months is the market’s own valuation of what was handed over. It is worth more than any auditor’s estimate, because somebody actually paid it.
4.3 — The auction that supplied the comparison
In April and May 2010 the government auctioned a different band of spectrum, for third-generation services, by open bidding.
It raised around ₹67,700 crore for the 3G bands, and about ₹1.05 lakh crore in total when the parallel broadband auction is included. The reserve price was exceeded several times over.
That auction is where the ₹1.76 lakh crore came from. The CAG took the price per unit of spectrum that operators had actually bid in 2010 and applied it to what had been given away in January 2008.
This is why the figure had such force. It was not an economist’s model. It was the government’s own selling price for a comparable asset, obtained eighteen months later from the same companies.
It is also where the objection lives. The 2010 auction was for a different band, for a different technology, in a market that had added roughly a hundred and fifty million subscribers in between. Whether a price obtained in one is the right price for the other is exactly the judgement Chapter Three said a presumptive loss rests on.
Nothing in this chapter turns on which view of that is right. What the 3G auction establishes beyond argument is narrower and quite sufficient: the government knew, in May 2010, what spectrum fetched when it was auctioned, and it had given a comparable quantity away for a fraction of that twenty-eight months earlier.
4.4 — What the audit found, and what the Supreme Court did
The CAG reported in November 2010. The figure of one lakh seventy-six thousand crore rupees entered the language. Chapter Three has already set out what it is and what it is not.
The Supreme Court’s judgment came on the second of February 2012. It cancelled all one hundred and twenty-two licences.
The reasoning is the part worth having, because it is the finding that has never been reversed, and it rests on a principle worth naming.
The idea that certain things — air, water, the seashore, minerals, radio frequencies — are not owned by a government in the way a desk is. The state holds them on behalf of the people, and must deal with them for the common benefit.
The practical consequence is that a government cannot simply give such a thing away on whatever terms it likes, even lawfully and even with good intentions. It has to be able to explain why the terms served the public.
This is the principle both the 2012 spectrum judgment and the 2014 coal judgment turn on, and it is the most consequential legal idea in this part.
Applying that principle, the Court reached three conclusions.
Court-found. The Court held that:
Spectrum is a natural resource held by the state in trust for the people, and its allocation must serve the common good.
The first-come-first-served method, as operated here, was arbitrary — because the alteration of the cut-off date and the manner of the January 2008 process meant that what determined who got a licence was not merit or price but who could reach a counter with a draft.
All one hundred and twenty-two licences were quashed.
This finding stands today. Nothing that happened afterwards has disturbed it.
4.5 — The criminal case
The CBI charged A. Raja, the DMK MP Kanimozhi, the former Telecom Secretary, Raja’s private secretary, and executives of several companies. A separate money-laundering case was brought by the Enforcement Directorate, alleging that ₹200 crore had been paid to a television channel connected to the DMK leadership.
Raja spent about fifteen months in custody before being bailed. So did several others.
The trial ran for six years before a special judge appointed to hear nothing else.
On the twenty-first of December 2017, Special Judge O. P. Saini acquitted every accused in all three cases.
Almost everybody reading about that acquittal assumed one of two things: either that it proved nothing had happened, or that it proved the courts had been got at.
Both assumptions share a hidden premise — that the criminal verdict is the final word on the whole affair. It is not. It is the answer to one of the two questions in Chapter Three’s second table.
The Supreme Court’s 2012 finding — that the allocation was arbitrary and that the licences must be cancelled — was not appealed, not reversed, and is not affected by the acquittal. It stands. The criminal court was asked a completely different question: whether the prosecution had proved, beyond reasonable doubt, that these named individuals had committed offences.
The judge’s answer was that it had not, and his stated reason was not that the allocation was proper. It was that the prosecution had failed to produce evidence, had misread its own material, and had built a case on inference.
So the honest summary is: the policy was found unlawful and the people were not found guilty. Those are compatible, they are both on the record, and neither side of Indian politics has ever wanted to say both sentences together.
4.6 — What happened to the industry
Cancelling a hundred and twenty-two licences in February 2012 was not a paper exercise. Companies had built networks on them and customers were using them.
The foreign operators that had bought into the Indian market on the strength of those licences — a Norwegian group, a Gulf group, a Russian one — found their businesses extinguished by a court judgment about a process they had not run. Several wrote off very large investments and left the country. Their Indian subscribers had to be migrated or lost.
India had around a dozen mobile operators in 2010. It now has, in effect, three, one of which is state-owned and struggling.
Not all of that is the 2012 judgment. A price war after 2016 did most of the rest. But the cancellations began it: they removed the newest entrants, wrote off the capital of the foreign companies that had funded them, and made India a market that outside telecom investors have largely avoided since.
The public got its money — ₹5.64 lakh crore from spectrum auctions between 2010 and 2024 — and lost most of its competition. Whether that trade was worth making is a real question and nobody has answered it, because it has never been asked.
Which is where the best defence of what was done in 2008 actually lives, and it is not the one its defenders usually make.
The defence of the original policy, at its strongest, is here rather than in the price. The reason for administered pricing was to get many operators into the market quickly, on the reasoning that competition drives tariffs down faster than any regulator can.
It worked. Indian call rates in that period became among the cheapest in the world, and the subscriber base grew from around four million in 2001 to over nine hundred million. A policy that achieves its stated object at that scale is not obviously a scandal.
Where it fails, and it is the same place as before: nothing in that argument requires a moved cut-off date, a counter, or a queue. Cheap spectrum could have been given cheaply to every qualified applicant. The Supreme Court cancelled the licences over how they were handed out, not over what was charged, and the industry consequences above flow from that method rather than from the price.
4.7 — What happened to the acquittal
The CBI and the Enforcement Directorate both appealed to the Delhi High Court in March 2018.
The appeal was not admitted until March 2024.
Six years and roughly thirty hearings, before nine different judges, to decide the preliminary question of whether the appeal could be heard at all.
Leave to appeal is a threshold step. It is not the appeal. In March 2024 the Delhi High Court granted it, and the substantive appeal is still pending.
By then A. Raja had been re-elected to the Lok Sabha.
Part Fourteen found telecom convictions taking fifteen years and urea convictions taking twenty-three, and concluded that delay is not a failure of the anti-corruption process but is the process. Here the same clock is running on the other side: an acquittal that the state says is wrong has stood unexamined for as long as some people’s entire careers.
The defence of what was done here is better than its reputation and has to be put at full strength.
The defence says: a court sat for six years, heard the whole prosecution case and acquitted everybody, saying there was no evidence of criminality. That is what an acquittal means. Continuing to describe this as a scandal after that is exactly what this book condemns when it is done to Congress figures elsewhere.
It adds a policy point that is genuinely strong: cheap spectrum was deliberate policy, pursued by governments of both parties, on the reasoning that Indians would get telephones faster if operators were not made to pay auction prices first. India’s mobile call rates became among the cheapest in the world. Auditing that policy against an auction that nobody intended to hold is auditing a choice, not a crime.
Where the defence wins: on the criminal charge, as of today, entirely. And the policy argument for administered pricing is real and was held by serious people.
Where it fails: the policy defence explains a low price. It does not explain changing the cut-off date, or a first-come-first-served counter, or why the companies that arrived first included two with no telecom operations that sold stakes within months. A policy of cheap spectrum could have been executed by giving it cheaply to everybody who qualified. The Supreme Court cancelled the licences over the method, not the price, and that finding is untouched.
Holding both of those in mind at once is the whole discipline of this part, and almost nobody in Indian public life does it.
January 2008: 122 licences at 2001 prices, first-come-first-served, after the cut-off date was moved. Two winners had no telecom operations and sold stakes within months.
February 2012: the Supreme Court cancelled all 122 and called the method arbitrary. That finding stands.
December 2017: a criminal court acquitted everybody, saying the prosecution had produced no evidence. The appeal took six years just to be admitted and is still pending.
The policy was found unlawful and the people were not found guilty. Both sentences are true.
5Coal
Coal is the other half of this part and it is bigger, older and less well understood than spectrum. The audit figure attached to it was larger. The Supreme Court’s finding was broader. And the period it covers is not five years but twenty-one.
5.1 — How coal blocks were given away
Coal was nationalised in the 1970s. From 1993 the law was changed to let private companies mine coal for their own use — a steel plant mining its own coal, a power station mining its own fuel. Those are called captive blocks.
The blocks were allocated by a screening committee: officials meeting, considering applications, and deciding who should get which seam. No money changed hands for the block itself. There was no auction.
A captive block is a coal deposit given to a company to mine for its own factory rather than to sell. The condition is that the coal must be used in the end-use plant.
A screening committee is a group of officials who decide who gets what. There is no bidding. Applicants state their case and the committee chooses.
The entire question of this chapter is whether that method — chosen in 1993, kept until 2010 — was defensible, and if it was not, for how long it had not been.
5.2 — What the audit found
The CAG’s final report went to Parliament in August 2012. Its figure was one lakh eighty-six thousand crore rupees. Chapter Three has already dealt with the draft figure of ten point six seven lakh crore that leaked five months earlier.
The substantive finding was narrower than the number and much harder to answer.
The government had itself concluded, by 2004, that competitive bidding was the right method. A legal opinion was sought. Draft rules were prepared. And then nothing happened for six years, while allocations by screening committee continued. Competitive bidding was legislated in 2010.
The delay between deciding to auction coal blocks and actually doing it was about six years.
In those six years a large number of blocks were allocated by the old method — to companies that, in a competitive process, would have had to pay.
This is the charge that does not depend on the presumptive loss figure at all. It does not require you to accept any estimate of what an auction would have raised. It requires only the government’s own conclusion that auctions were correct, and the six years it then took.
Because everything else in this chapter can be argued about, it matters where that particular fact comes from.
The six-year delay is not an inference from the audit’s loss figure. It is documented in the ministry’s own file.
The record shows the Ministry of Coal concluding that competitive bidding was the right route, seeking a legal opinion on whether it required a change in the statute, receiving that opinion, preparing draft rules — and then the matter sitting while the screening committee carried on allocating.
Those files are public because of the transparency law of 2005. The strongest single charge in this part exists in citable form because of a statute this same government passed, which is the point Chapter One made and Chapter Nine will make again from the other side.
Nobody has ever disputed the sequence. What is disputed is what it means — whether it shows obstruction, or a slow ministry, or a genuine legal difficulty about whether auctions needed primary legislation.
5.3 — What the Supreme Court found, and how far back it reached
On the twenty-fifth of August 2014 the Supreme Court held that the allocation of coal blocks had been arbitrary and illegal. On the twenty-fourth of September it cancelled two hundred and fourteen of the two hundred and eighteen blocks allocated, imposed a levy of ₹295 per tonne on coal already extracted from them, and allowed producing mines a transition period to the end of March 2015.
Now read the date range in that judgment carefully, because it is the thing this chapter is actually about.
Everybody calls this the UPA coal scandal. The audit was of the UPA years, the political fight was about the UPA, and the Prime Minister held the coal portfolio between 2006 and 2009.
But the Supreme Court declared the allocations illegal from 1993 onwards, and cancelled blocks allocated across twenty-one years and five different governments — Congress, United Front, and the National Democratic Alliance.
The screening committee method was created in 1993, under the government of Part Fourteen. It was used by every government that followed. The mechanism the Supreme Court struck down was twenty-one years old and had been operated by everybody.
Part Ten found that not one instrument used in Punjab was a Punjab instrument. Part Nine found that every instrument of the Emergency existed before it. This is the same finding in a third setting, and it is the most-repeated finding in this entire series: the thing that gets called a scandal is usually a machine that was already running.
Which cuts hard in both directions here. It means the charge against this government is narrower than the name suggests — and it means the charge against the party as a continuous actor, which is what this series charges, is very much wider. The party that created the method in 1993 is the party that had not replaced it by 2010.
5.4 — What happened to the blocks
Cancelling two hundred and fourteen coal blocks in September 2014 created an immediate problem, because steel plants and power stations had been built on the assumption that the coal underneath them belonged to somebody.
The Court had allowed producing mines to run to the end of March 2015 and imposed a levy of ₹295 per tonne on coal already extracted. That left about six months to build a new legal basis for the entire captive coal sector.
It was done by ordinance in October 2014 and then by statute in March 2015, creating the auction mechanism under which the blocks were re-sold. Chapter Eight sets out what those auctions produced and what the figures actually meant.
Between the judgment and the re-auction, companies with plants already built were buying coal on the open market or importing it, at prices well above what mining their own block would have cost.
Several of the re-auctioned blocks were bought back by the same companies that had lost them — this time at a price, and in some cases at a price high enough that the economics of the plant no longer worked. Six auctioned blocks were subsequently cancelled again for missing development deadlines.
The public asset was recovered and re-sold properly. The plants built on the old basis were a separate casualty, and nobody has ever added up what that cost.
That is a smaller version of the calculation Chapter Ten asks for, and it has not been done either.
One consequence of the 2015 statute belongs on the record because it answers the largest structural question in this chapter. From 2015 the default for allocating a coal block is an auction. The screening committee is gone.
The method that ran from 1993 to 2010, that no government replaced in seventeen years, and that this government legislated against in 2010 without ending, was finally ended by a Supreme Court judgment and six months of emergency drafting. Every one of Chapter Nine’s rules exists because the alternative to a rule is that a court eventually does it instead, badly and in a hurry.
5.5 — The criminal cases
Coal produced something spectrum did not: convictions.
In March 2016 two directors of a Jharkhand company were sentenced to four years. In 2017 the former Coal Secretary, H. C. Gupta, and two other officials of the coal ministry were convicted and sentenced to two years each over a Madhya Pradesh block, with a company managing director given three years and the company fined a crore. They were the first public servants sentenced in these cases.
Gupta faced around eight separate charge sheets, tried individually after the Supreme Court declined to allow a joint trial.
Several of those cases ended the other way. In March 2026 the former MP Vijay Darda and H. C. Gupta were acquitted in the Bander block case, which closed the oldest pending coal matter.
The defence of the officials is worth stating because it is unusual and because it was made by the convicted man himself.
Gupta argued that the audit’s method was technically wrong — that it applied a single average mining cost across blocks whose real costs varied enormously with the depth of the coal, the amount of earth to be removed, the grade, and the distance to a railhead. Averaging those, he argued, produced an inflated gain figure and therefore an inflated loss.
He argued further that the blanket cancellation of two hundred and fourteen blocks in 2014 destroyed viable projects along with indefensible ones, because the Court struck down a method rather than examining allocations case by case.
And he made a point about consequences that this series should record: that officials who applied a policy they did not set were prosecuted individually, over years, in separate trials, while nobody who set the policy was.
Where this fails: the technical criticism of the averaging goes to the size of the number, and Chapter Three has already conceded that the number is soft. It does not touch the finding that the method was arbitrary.
Where it lands: squarely. A Coal Secretary receiving a two-year sentence for operating a screening committee created in 1993, while the six-year delay in replacing it produced no consequence for anybody at all, is a real description of what happened.
That last point is the one this book takes forward into Chapter Nine, because the remedy for it is a rule and the rule was never made.
Coal blocks were given out by a committee of officials, with no auction, from 1993. The government concluded by 2004 that auctions were right and legislated them in 2010 — a six-year gap in which allocations continued.
In 2014 the Supreme Court called the allocations arbitrary and illegal since 1993 and cancelled 214 of 218 blocks — across five governments and three parties.
The people who went to prison were mostly civil servants who operated the method. Nobody was punished for the six years.
6The Games, and the Building
Two smaller cases, taken together because they are the same case in different clothes: public value transferred through procurement in one, and through land in the other.
6.1 — Delhi, October 2010
India hosted the Commonwealth Games in Delhi in October 2010. The preparation was chaotic and publicly so — a collapsed footbridge, unfinished accommodation, and delegations complaining before the opening.
The CAG report ran to seven hundred and forty-three pages. Its findings were about procurement: contracts awarded at inflated rates, competitors disqualified so that a single tender remained, bids evaluated after the award had effectively been decided, and costs multiplying against the original estimates.
A committee under V. K. Shunglu was appointed by the Prime Minister and reported separately.
The timing of the audit is what makes it unusually strong. Parts of the CAG’s findings were known before the Games were held, not years afterwards, because the audit ran alongside the preparations.
That means the material was not reconstructed from memory by people defending themselves. Contracts, tender documents and evaluation records were examined while the work was still being done.
The report indicted the organising committee, the Delhi government and the Prime Minister’s Office for their respective parts, which is a wider circle than the political argument at the time — which was almost entirely about one man.
The scale of what was being audited is worth stating before the cases are traced, because the sum involved is often quoted without saying what it covered.
The Games were bid for in 2003 on an estimate in the region of a few hundred crore rupees for the organising committee’s own budget.
The total public spending associated with them — venues, the athletes’ village, city works, the organising committee — is generally put in the region of ₹70,000 crore. The CAG’s report found the final costs of individual items running at multiples of what had been approved.
Much of that was city infrastructure that Delhi has used ever since, and it is not honest to count a metro extension as money lost. The audit’s charge was not that the money was spent. It was how the contracts to spend it were awarded.
6.2 — What happened to it
Suresh Kalmadi, chairman of the organising committee and a Congress MP, was removed from his post in January 2011, arrested in April, and held in Tihar jail for around nine months before bail. He was suspended by his party.
Around nineteen First Information Reports were registered.
When an investigating agency concludes that the evidence does not support a prosecution, it does not simply stop. It files a closure report asking the court to end the case.
The court is not obliged to agree. It can reject the report and order further investigation, as happened in Part Fourteen’s urea case, where a closure report filed after twenty-two years was thrown out.
A closure report accepted by a court is therefore a judicial finding of a kind — not that the accused is innocent, but that the state, having investigated, cannot make a case.
Then, over the following fifteen years, almost all of it dissolved.
| When | What happened |
|---|---|
| 2010–11 | CAG report; Shunglu committee; ~19 FIRs; Kalmadi arrested and jailed nine months. |
| Jan 2014 | CBI files a closure report on the two main contracts, citing lack of incriminating evidence. |
| Feb 2016 | A special court accepts that closure report. |
| Apr 2025 | A special court accepts the Enforcement Directorate’s closure report in the money-laundering case, on the ground that no scheduled offence had been committed and there were therefore no proceeds of crime. |
| To date | At least two convictions in smaller matters, with appeals pending in the Delhi High Court and the convicts on bail. |
The Prime Minister’s Office declined an application under the transparency law for details of action taken on the Shunglu report, on the ground that a group of ministers was still considering it.
Nineteen FIRs. Fifteen years. Nobody has been held responsible for the central allegations.
The man at the centre of it spent about nine months in prison in 2011 on charges that were, in the principal case, formally abandoned by the prosecuting agency in 2014 and by the money-laundering agency in 2025.
He was eighty when the last case ended. He was never convicted of anything in it, and he spent nine months in Tihar. Both halves of that are failures of the same system, and they point in opposite directions.
6.3 — What was left behind
One thing has to be said for the Games and it is regularly left out of accounts hostile to this government.
They happened. The venues were finished, the events were held, and India’s athletes had their most successful Commonwealth Games. A large amount of Delhi’s transport infrastructure — metro extensions, roads, an airport terminal — was built or accelerated for them and has been in daily use since.
Delhi’s metro carries several million passengers a day. Terminal 3 at the airport, opened in 2010, handles a large share of India’s international traffic.
Both were built to a deadline the Games created.
The audit’s charge was never that the money was wasted in the sense of producing nothing. It was that the contracts to produce it were awarded by rigged tendering at inflated rates — that a thing worth building was bought badly.
Those are different charges and only the second one is made here. Several of the sporting venues, by contrast, have been substantially unused since, which is the ordinary fate of stadiums built for a fortnight.
6.4 — Colaba
The Adarsh Co-operative Housing Society was a thirty-one storey block built on government land in Colaba, in the most expensive part of Mumbai. The land had been intended for accommodation connected with defence personnel and, in the account that circulated at the time, for the widows of the Kargil conflict.
In November 2010 it emerged that flats had gone to serving and retired officials, bureaucrats and politicians, including relatives of the Chief Minister of Maharashtra, Ashok Chavan. He resigned within days.
A two-member judicial commission of Justices J. A. Patil and N. N. Kumbhar examined the matter and reported in 2013 after hearing a hundred and eighty-two witnesses.
Commission-found. Of the hundred and two members of the society, twenty-five were ineligible. There were twenty-two cases of flats held through proxies.
The commission found instances of impropriety and corruption in the allotments against a number of senior officials and politicians. It found that clearances and permissions had been given as a quid pro quo.
It did not find the then Chief Minister singularly or particularly culpable.
The CAG had reported in 2011 that the episode showed how a group of officials in key posts could subvert rules to obtain prime public land for personal benefit.
What followed is a study in how prosecution sanction works, and it is worth setting out in order because it is the clearest example in this series.
| When | What happened |
|---|---|
| 2012 | CBI charge sheets thirteen people, including the former Chief Minister. |
| 2013–14 | The Governor refuses sanction to prosecute him, citing insufficient evidence. The CBI applies to drop his name; the trial court refuses. |
| Feb 2016 | A different Governor grants sanction on the basis of what the CBI says is fresh evidence. |
| Apr 2016 | The Bombay High Court orders the building demolished. |
| Dec 2017 | The Bombay High Court quashes the sanction, holding that the CBI had failed to present any fresh evidence and that material already considered could not be converted into fresh material. |
Two Governors, appointed by governments of different parties, reached opposite conclusions on the same file. The High Court then held that the second had been given nothing new to act on.
On the war widows. The version that circulated in 2010 — that flats meant for the widows of Kargil had been taken by officials — is the most damaging sentence attached to this case, and it is the one part of it this book cannot stand behind.
The society’s stated purpose changed several times across a decade of correspondence, and the proposition that a fixed allocation for war widows existed and was diverted is Alleged rather than established. It is repeated everywhere, including by people who ought to check.
Part One’s chapter on fabrication makes the strategic argument, and this is a textbook case of it: the widows sentence is what lets every documented finding in this section be waved away as more of the same. Drop it and what remains is stronger.
On the former Chief Minister. The judicial commission that examined the matter over 182 witnesses did not find him singularly or particularly culpable. His relatives’ flats were returned. The Bombay High Court found the CBI had produced no fresh evidence when sanction was granted in 2016. He has never been convicted of anything.
What survives all of that is the commission’s actual findings: a quarter of the members ineligible, twenty-two proxy holdings, and clearances given as a quid pro quo, on public land in the most expensive part of Mumbai. Those depend on neither the widows, nor the Chief Minister, nor any prosecution.
Prosecution sanction is usually described as a safeguard — a check that stops officials being harassed with vexatious cases for doing their jobs. Look at what it did here. The same file produced a refusal under one Governor and a grant under another, three years apart, with a change of government at the Centre in between, and the High Court then found that nothing new had been added.
So the requirement did not filter the evidence. It filtered nothing at all: the material was identical both times, and what changed was who was asked.
This series has now met that device three times. Part Thirteen found Section 6 of the Armed Forces (Special Powers) Act barring prosecution without central sanction, and sanction refused over the Nagaland killings of December 2021 even after the state cabinet asked for it. Part Fifteen found the Single Directive struck down in 1997 for the same reason. A sanction requirement does not test a case. It relocates the decision to somebody political — and here it shielded a Congress Chief Minister under one Governor and was turned against him under another. An instrument that can do both is not a safeguard; it is a switch.
One further fact belongs on the record, in the interest of the evenhandedness this series claims. The beneficiaries of Adarsh allotments were not confined to one party. Figures who later held office in the National Democratic Alliance were also named as allottees. The commission’s findings were about a system of allotment, not about a single party’s people.
The Games: nineteen FIRs, fifteen years, and closure reports accepted in 2016 and 2025. The man at the centre spent nine months in Tihar and was never convicted in the main case.
Adarsh: a judicial commission found 25 of 102 members ineligible and 22 flats held through proxies. A Chief Minister resigned.
Then one Governor refused permission to prosecute him, another granted it three years later, and the High Court held nothing new had been produced in between. The evidence never changed. Only the person being asked did.
7The Parliament That Stopped
Something happened to Indian government between 2010 and 2014 that is usually described as a mood and was in fact a measurable failure of an institution.
7.1 — The numbers
The fifteenth Lok Sabha, sitting from 2009 to 2014, passed 179 bills out of 291 introduced — about fifty-four per cent, the lowest rate of any full-term Lok Sabha in Indian history.
More than thirty per cent of scheduled sitting time was lost to disruption. In the years from 2011 to 2013 productivity fell below half.
Only about thirteen per cent of the time the House did function was spent on legislative scrutiny.
An entire session in the winter of 2010 was lost almost completely to a demand for a joint parliamentary committee on spectrum.
The government’s response to a legislature that would not sit was to legislate less, and to govern by other means where it could. The phrase used at the time was policy paralysis. It described something real: investment decisions deferred, clearances delayed, and ministers reluctant to sign anything after watching a Coal Secretary prosecuted for signing things.
7.2 — What happened outside the building
While Parliament stopped working, something took its place, and this part cannot be understood without it.
In April 2011 Anna Hazare, a social activist from Maharashtra, began a hunger strike in Delhi demanding a strong anti-corruption ombudsman. He fasted again in August, for twelve days, at the Ramlila Maidan. The crowds were very large and the coverage was continuous. The government arrested him, released him within hours, and then negotiated.
In December 2011 the Lokpal Bill passed the Lok Sabha and stalled in the Rajya Sabha. It did not become law for another two years.
An ombudsman: an independent body empowered to investigate corruption complaints against public officials, including ministers and, under the 2013 Act, the Prime Minister.
The idea is not new. A Lokpal bill was first introduced in 1968. Versions were introduced in 1971, 1977, 1985, 1989, 1996, 1998, 2001 and 2008. None passed.
Forty-five years, ten attempts, under governments of every description. That is the background against which the 2011 movement should be read.
Two things came out of those two years and they point in opposite directions.
The Lokpal Act was passed in December 2013, forty-five years after the first attempt. It is a real statute and it is on the books.
The first Lokpal was appointed in March 2019 — five years and three months later, under a different government, and after the Supreme Court had been asked twice to compel an appointment.
The other thing that came out of it was a political party. The Aam Aadmi Party was formed in November 2012 by people from the movement, won twenty-eight seats in Delhi in December 2013, and formed a government there.
A mass movement demanding a law produced the law, a party, and an office that stood empty for five years. Which of those three counts as the result is a matter of judgement, and this book does not think it is the first.
There was a serious case for resisting what the movement demanded, and it was not only a self-interested one.
The defence of the government’s handling says: a hunger strike is not a legislative procedure. A movement that demands a specific text and refuses amendment is asking Parliament to abdicate, and several serious constitutional lawyers said so at the time — the draft the movement pressed would have created a body with investigative, prosecutorial and adjudicatory powers combined, answerable to nobody.
A government that resisted that and passed a different version in 2013 was doing its job, not obstructing.
Where this wins: the objection to combining those powers in one body is a real constitutional objection and was not invented for the occasion.
Where it fails: the government’s problem was not that it opposed the movement’s draft. It was that it had had forty-three years and eight bills before the movement started, and produced nothing. A party that has declined to legislate on a subject since 1968 has forfeited the argument that it is being rushed.
7.3 — Whose fault the House was
It is usual to say that a Parliament which does not function is being obstructed, and that the fault lies with whoever is doing the shouting. That will not survive a look at the record. Disruption as a tactic was not invented in 2010: the Congress used it in opposition before 2004 and after 2014, and the parties that used it against this government have been on the receiving end since. It is not a party’s habit but a property of the rules.
An Indian opposition has almost no procedural instrument that compels a government to answer. It cannot force a select committee, cannot compel a minister to attend, and cannot summon papers. What it can do is stop the House, and that is the only lever with any weight. Disruption is therefore not a failure of the institution’s rules but the predictable output of them.
Part Eight asked the reader to name an occasion in nine years when Parliament stopped a government, and found one, lasting a day. This is the mirror image: a Parliament with no way of scrutinising a government and one way of stopping it, using the one it has.
None of which excuses the government of anything. It had a working majority in the lower house for five years and passed the smallest proportion of its own bills of any full-term government in the country’s history. Part of that is disruption. Part of it is that after 2011 it stopped trying.
7.4 — What it cost, measured
“Policy paralysis” was a phrase, and phrases are cheap. What it referred to can be measured.
Growth ran at around eight per cent in the first years of this government and fell to roughly four and a half per cent by 2012–13 — close to half.
Inflation, particularly in food, ran in double digits for long stretches.
In the summer of 2013 the rupee fell sharply against the dollar, from about fifty-four to nearly sixty-nine in a few months, as foreign money left. The Reserve Bank had to take emergency measures.
India was described that year, by an outside bank, as one of the “fragile five” — a list of economies most exposed to a sudden withdrawal of foreign capital. Twenty-two years after the crisis of Part Fourteen, the same vulnerability was being discussed in the same language.
How much of that was caused by what is genuinely disputed, and this book will not pretend to settle it. Global conditions turned after 2011. The American central bank signalled a change in 2013 that moved money out of every emerging market, not only India.
What is not disputed is the part that was domestic: clearances stalled, projects held up, and — the specific mechanism this part is about — officials who had watched a Coal Secretary prosecuted for signing an allocation became extremely reluctant to sign anything.
Everybody assumes that prosecuting corrupt officials reduces corruption. It is such an obvious proposition that stating it feels unnecessary.
But watch what actually happened. A Coal Secretary was convicted for operating a screening committee created in 1993, in a case where nobody alleged he took money. The message received across the civil service was not do not take bribes. It was do not decide anything.
A decision that is never made cannot be prosecuted. A file that sits produces no charge sheet. So the rational response to a prosecution for exercising discretion is to stop exercising discretion — which is exactly what a large part of the Indian administration did between 2012 and 2014, and to a degree has not stopped doing.
Anti-corruption enforcement that cannot distinguish a bad decision from a corrupt one does not produce honest decisions. It produces no decisions.
Part Four found that a permission system is worth more to the incumbent than to the regulator. This is the same shape one layer up: an accountability system that punishes judgement rather than dishonesty makes paralysis the safest career. And nothing in the rules has changed since.
7.5 — The ordinance
One episode from this period belongs in the record because it says more about the arrangement described in Part Fifteen than anything else in either part.
In July 2013 the Supreme Court held, in a case brought by the lawyer Lily Thomas, that a Member of Parliament or of a state legislature convicted of an offence carrying a sentence of two years or more is disqualified immediately, and struck down the provision that had allowed a convicted legislator to stay in office while appealing.
In September 2013 the government brought an ordinance to undo that.
A law made by the executive when Parliament is not sitting, on the President’s signature. It has the full force of an Act immediately.
It lapses within six weeks of Parliament reassembling unless the House passes it. So an ordinance is a way of legislating first and asking afterwards.
The power exists for genuine emergencies. Chapter Seven has already described a Parliament that lost thirty per cent of its sitting time, and a government that cannot get its business through a House has an obvious temptation to use it instead.
On the twenty-seventh of September, at a press conference, the Congress vice-president Rahul Gandhi described the ordinance as complete nonsense and said it should be torn up. The Prime Minister was abroad. The ordinance was withdrawn.
The defence says: the outcome was right. A bad ordinance that would have protected convicted legislators was killed, publicly, by the ruling party’s own leadership. The disqualification rule stands to this day and has removed legislators of every party since. Judge the result.
The reply: the result was right and the method was the problem this whole part has been describing. The ordinance had been approved by the Cabinet. It was killed not in Parliament, not by the Cabinet reconsidering, and not by the Prime Minister, but at a press conference by a person holding no government office, while the Prime Minister was out of the country.
Part Fifteen’s opening box asked what happens when the office of Prime Minister is separated from the authority that goes with it. This is the answer, on camera, in public, and it is the single clearest illustration in nineteen parts.
7.6 — How it ended
In May 2014 the Congress won forty-four seats. The alliance won fifty-nine. It was the worst result in the party’s history and remains so.
From 206 seats to 44 in five years. The party fell below the threshold conventionally required to claim the office of Leader of the Opposition in the Lok Sabha.
Part Fourteen recorded a Congress government reduced to 140 after rescuing the economy from default. Part Fifteen recorded one increasing its majority by sixty-one seats after four years of bombings.
Three consecutive Congress governments; three electoral verdicts; and in none of the three does the verdict track the record in any consistent way. Chapter Nine draws the conclusion, and it is uncomfortable for a book like this one.
What that inconsistency means for a book which assumes that records and verdicts are connected is the subject of the next chapter but one.
The 2009–2014 Lok Sabha passed 54% of the bills introduced in it — the lowest of any full-term Parliament in Indian history — and lost over 30% of its sitting time to disruption.
Disruption is not a party’s bad habit. It is the only instrument the rules give an opposition, and every party uses it when out of power.
In September 2013 a Cabinet-approved ordinance was killed at a press conference by a man holding no government office, while the Prime Minister was abroad.
8What Was Actually Established
This is the chapter Part One promised. It sets out, case by case, what stands today and what does not.
Read it with Chapter Three’s second table in mind. Two questions, always: was the allocation proper, and did a named person commit a crime.
8.1 — The reckoning
| Case | The allocation question | The criminal question |
|---|---|---|
| Spectrum, 2008 | Court-found — arbitrary; 122 licences cancelled by the Supreme Court, February 2012. Never disturbed. | All accused acquitted, December 2017. Appeal admitted March 2024, still pending. |
| Coal, 1993–2010 | Court-found — arbitrary and illegal since 1993; 214 of 218 blocks cancelled, September 2014. Never disturbed. | Mixed. Convictions of officials and company directors from 2016; several later acquittals, the last in March 2026. |
| Commonwealth Games | Audited — CAG found inflated contracts and rigged tendering. No constitutional court ruled. | Closure reports accepted, 2016 and 2025. Two convictions in minor matters, under appeal. |
| Adarsh | Commission-found — 25 of 102 members ineligible, 22 proxy holdings, quid pro quo clearances. Building ordered demolished, 2016. | Prosecution sanction against the former Chief Minister quashed, December 2017. Other proceedings continuing. |
The left column is uniformly against the government. The right column is uniformly not proved, with the partial exception of coal, where those convicted were mostly officials rather than ministers.
8.2 — The test that was actually run
Chapter Three said a presumptive loss compares a real transaction with an imagined one, and that the figure depends entirely on the comparison chosen. That sounds like an argument nobody can settle.
But it was settled, at least partly, because both assets were subsequently auctioned for real. What follows is the only evidence in this part that tests the audit figures against money that actually changed hands.
Spectrum
After the Supreme Court cancelled the licences, the spectrum had to be re-sold. The first auction was held in November 2012.
The November 2012 auction failed. Much of the spectrum went unsold. The government had budgeted around ₹58,000 crore from spectrum for that year and ended up expecting about ₹19,400 crore.
The Finance Minister, P. Chidambaram, said publicly that the ₹1.76 lakh crore figure had been “a pure myth”, and predicted the coal figure would prove equally fanciful.
For about two years, that looked like the end of the argument.
Then it reversed. The auction of February 2014 raised around ₹62,000 crore. The auction of March 2015 raised about ₹1.10 lakh crore. Between 2010 and 2024, spectrum auctions have realised roughly ₹5.64 lakh crore in total.
Coal
The cancelled blocks were re-auctioned from February 2015. The government announced in March 2015 that proceeds from the first tranches — combining auction receipts, royalties and upfront payments over the life of the mines — had crossed ₹2 lakh crore, and said explicitly that this exceeded the CAG’s ₹1.86 lakh crore estimate. The Prime Minister of the day noted that auctioning fewer than ten per cent of the cancelled mines had produced ₹1.10 lakh crore.
That reads as a complete vindication of the audit. It is not, and the reason is exact.
Everybody who cited the 2015 coal auctions assumed that the auction figure was a real number in a way the audit figure was not. One was money; the other was an estimate.
Look at what the ₹2.07 lakh crore consisted of. It was e-auction proceeds plus royalties plus upfront payments, projected across the working life of the mines — a period given as around thirty years. The government’s own estimate was that ₹3.35 lakh crore would flow to the states over three decades.
By July 2018 — three and a half years in — the amount actually collected and passed to the states was about ₹5,700 crore. Against an average implied by the projection of roughly ₹11,000 crore a year, the collection over three and a half years came to about a seventh of it. Six auctioned blocks had been cancelled for missing deadlines.
So the number used to prove that the presumptive figure was real was itself a presumptive figure — a projection of receipts over thirty years, quoted as though it were a bank balance, by the government that had campaigned against presumptive figures.
Chapter Three said this technique is legitimate and that its output depends on the comparison chosen. Here is the proof, and it is the reverse of the one anybody wanted: both sides of Indian politics used a presumptive loss when it suited them and called it a myth when it did not, and neither has ever gone back to check.
What the test actually shows
| Asset | The audit said | What auctions then produced | Verdict on the figure |
|---|---|---|---|
| Spectrum | Up to ₹1.76 lakh crore forgone | Nov 2012 auction largely failed; Feb 2014 ~₹62,000 cr; Mar 2015 ~₹1.10 lakh cr; ~₹5.64 lakh cr cumulative to 2024 | Broadly supported over time, badly wrong on timing. Spectrum was worth a great deal — but not in 2012. |
| Coal | ₹1.86 lakh crore forgone | ₹2.07 lakh cr claimed in 2015 — but that was a 30-year projection; ~₹5,700 cr actually collected by July 2018 | Untested. The figure said to vindicate it was another presumptive estimate. |
This is the nearest thing to an answer that exists, and it is partial. The full comparison Chapter Ten calls for — every cancelled block, every re-auction, actual receipts against the 2012 estimate — has still never been done by anybody.
The honest conclusion is unsatisfying and it is the one this part will stand on: the spectrum figure was closer to right than its critics said and less right than its users claimed, and the coal figure remains genuinely unknown eleven years after the blocks were cancelled.
8.3 — What that pattern means
The pattern is too consistent to be an accident, and there are only three explanations available.
The first: nothing criminal happened. The allocations were bad policy, badly executed, and the criminal cases failed because there was no crime. On this reading the presumptive-loss figures were a category error that the press and the opposition turned into a decade of prosecution.
The second: crimes happened and the prosecutions failed. Cases built late, on inference, by agencies without the capacity to prove a bribe. On this reading the acquittals are about the investigators, not the accused.
The third: the cases were brought for reasons that had little to do with proving them.
The strongest evidence for the second explanation comes from the judges themselves.
The special judge in the spectrum case did not say the allocation was proper. He said the prosecution had produced no evidence of criminality, had misread its own material, and had built its case on conjecture. That is a finding about the CBI, not an endorsement of what was done in 2008.
The strongest evidence for the first is that the same conclusion was reached, independently, across four separate matters, several agencies and fifteen years — including under a government with every political reason to secure convictions.
This book cannot choose between them and does not pretend to. Chapter Ten lists it among the things not known, and it is the largest thing on that list.
Before setting out what does survive, the strongest objection to the way this chapter is built should be put on the page.
The strongest objection to this entire chapter: that splitting the allocation question from the criminal question is a lawyer’s distinction with no moral content. If a government hands a public asset to a company for a fraction of its worth, and that company sells a stake at a multiple within months, the public has been robbed. Whether a court can convict a named individual of an offence under a particular statute is a question about the law of evidence, not about what happened.
The reply, which concedes most of it: that is right about the harm. The public was worse off, the beneficiaries were better off, and a criminal acquittal does not reverse either. This book has said so in Chapter Three and repeats it in the verdict below.
Where the distinction still has to hold: because the alternative is a standard under which anybody can be described as a criminal on the strength of an outcome. Part Eleven refused to name individuals against whom no body had found. Part Twelve refused it for Rajiv Gandhi. Part Fifteen refused it for the officers at Batla House. A rule applied to protect people this book’s readers dislike is worth nothing unless it also protects people they do.
The harm is established. The crime is not. Saying both is not a compromise between them.
8.4 — What is not in doubt
Whichever explanation is right, a set of findings survives all three of them, and this is the charge that this part actually makes.
Established, and untouched by any acquittal:
That very large public assets were allocated by discretionary methods which two separate benches of the Supreme Court found arbitrary, in 2012 and 2014.
That in the coal case the government had itself concluded by 2004 that auctions were correct, and took six years to legislate them while allocating under the old method throughout.
That a judicial commission found a quarter of the members of a housing society on public land in Colaba ineligible, and clearances given as a quid pro quo.
That the audit of the Commonwealth Games found rigged tendering, and that fifteen years of investigation produced closure reports.
Not established, and this book does not assert it: that any minister of this government took money in any of these four matters. No court has found it. The one set of criminal trials that ran to completion on the largest case ended in acquittals of everybody.
That is the reckoning, and it is narrower than the reputation of these five years and wider than the acquittals alone would suggest.
Four cases. In every one, the allocation question went against the government — two Supreme Court judgments, a judicial commission, an audit.
In every one, the criminal question did not stick, except against officials in the coal cases.
Three explanations exist: no crime; crimes badly prosecuted; or cases brought for other reasons. This book cannot choose between them, and says so.
9What Would Have Prevented It
This chapter states the defence at full strength, says where it wins, and sets out the rules that would have changed the outcomes.
9.1 — The defence
One. The criminal charges failed. Every accused in the spectrum case was acquitted after a six-year trial by a judge appointed to hear nothing else. The Games cases ended in closure reports accepted by courts. The sanction to prosecute in Adarsh was quashed for want of fresh evidence. A book that treats a failed prosecution as a proved charge is doing what this series has refused to do for fifteen parts.
Two. The machinery was inherited. The coal screening committee dates from 1993. Administered pricing for spectrum was policy under governments of both parties. The Supreme Court struck down coal allocations across twenty-one years and five governments. This government did not build either method.
Three. The building record is large. Chapter Two. Education and food as rights, the 1894 land law replaced, the criminal law on sexual offences rewritten, a Lokpal legislated.
Four. The scandals are visible because this government made them visible. The transparency law of 2005 is how the file notings reached the public. A government that legislates the instrument which then destroys it has done something no earlier government in this series did.
Five. The presumptive-loss figures were not what they were taken to be, and the largest of them was five and a half times bigger in a leaked draft than in the published report. Chapter Three.
Two of those five come from bodies with no reason to help this government.
The acquittals and closure reports were delivered between 2014 and 2026, entirely under a government of a different party which had campaigned on these very cases. A CBI operating under that government filed the closure report in the Games matter in January 2014 and the Enforcement Directorate did the same in 2025.
And the Supreme Court’s coal judgment of 2014, which is the most damaging single finding in this part, is also the one that reached back to 1993 and implicated every government in between. The same judgment that convicts the party as a continuous actor acquits this particular government of having invented anything.
Six. It legislated the auctions. This is the point the defence rarely makes for itself, and it is the strongest one available.
Competitive bidding for coal blocks was put into law by an amendment in 2010 — before the CAG report, before the Supreme Court, and by this government. Spectrum was auctioned by this government in 2010 for the 3G bands, and it was that auction which supplied the comparison the audit then used against it.
The figure that destroyed this government politically was calculated from a price obtained in an auction it had itself held.
And the method whose absence the Supreme Court condemned in the coal judgment of 2014 had been legislated by the same government four years earlier.
Both charges therefore rest on the government’s own corrective acts. That is a real point in its favour and it does not appear in any account of these years.
It also contains the charge. A government that legislated auctions in 2010 had known they were right since 2004, and the six years in between are the whole case against it.
9.2 — Where the defence loses
On four things.
The six years on coal. This is the charge that survives every acquittal and does not depend on any audit figure. The government concluded internally by 2004 that competitive bidding was correct. It legislated it in 2010. In between it kept allocating by the method it had already decided was wrong. No court finding is needed for that and none has ever been contradicted.
The changed cut-off date. The policy defence of cheap spectrum explains a low price. It does not explain moving a cut-off date in January 2008 to exclude applicants who had applied in reliance on the published one, and it does not explain a counter and a queue. The Supreme Court cancelled the licences over the method.
The Lokpal. Legislated in December 2013 after two years of public pressure; no Lokpal appointed until March 2019, under a different government, after the Supreme Court was asked twice. A law is not a thing done until the office exists.
The communal violence law. Ten years, two terms, two majorities, four drafts, never brought to a vote. Part Fifteen set this out and Part Sixteen is where the clock ran out on it.
The paralysis. Chapter Seven measured it: the lowest bill passage rate of any full-term Lok Sabha, over thirty per cent of sitting time lost, growth halved, and an administration that stopped signing things. Part of that was an opposition using the one instrument the rules give it. Part of it was global. But a government with the strongest parliamentary position since 1984 that passed fifty-four per cent of its own bills has not been prevented from governing. It stopped.
And the specific mechanism is the one this book has to name, because it is uncomfortable for everybody. Officials stopped deciding after watching a Coal Secretary prosecuted for a decision in which nobody alleged he took money. That was a rational response to an accountability system that cannot tell a bad decision from a corrupt one, and no government of any party has fixed it since.
9.3 — The rules that were not made
| The failure | The rule that would have caught it | Status |
|---|---|---|
| Public natural resources allocated by discretion | A statutory default requiring auction of any finite natural resource, with departures reasoned in writing and laid before Parliament | Partly done. Auctions legislated for coal in 2010 and for spectrum after 2012 — by court compulsion, not by choice |
| Six years between deciding to auction and doing it | A statutory time limit between a government recording a policy decision and laying the implementing instrument, with a report to Parliament on expiry | Never proposed |
| A presumptive-loss figure treated as a measure of theft | A statutory requirement that audit reports state the counterfactual used and the range produced, on the face of the summary | Never enacted |
| Prosecution sanction turning on who is asked | Sanction decided by a body independent of the executive, with written reasons and an appeal | Never enacted; the same gap as Part Thirteen and Part Fifteen |
| A Lokpal legislated and not constituted | A statutory deadline for constituting any body created by an Act, with the default appointment made by the Chief Justice on expiry | Never enacted |
| Officials prosecuted for exercising discretion, with no allegation of gain | A statutory threshold requiring evidence of personal benefit or dishonest intent before a public servant is charged over a policy decision taken through proper channels | Partly attempted by an amendment in 2018; its effect is contested and it came four years too late for this part |
| Communal violence, no duty on officials | The bill drafted in 2005 and 2011 | Never passed, in ten years |
Parts Nine, Eleven, Thirteen, Fourteen and Fifteen each closed with a comparable table. Part Nineteen consolidates all of them, and the fourth and sixth rows here have now appeared three times each.
The first row of that table and the seventh are in tension, and it is worth saying so rather than leaving a reader to notice it.
A rule requiring public assets to be auctioned removes discretion from officials. A rule protecting officials who exercise discretion protects it. Both are on the list because the failure in this part was not too much discretion or too little — it was discretion with no record of reasons attached to it.
An official who allocates a coal block and writes down why, against stated criteria, laid before Parliament, is protected by that document if the decision is later questioned, and convicted by it if the reasons were false. An official who allocates a block with nothing on the file is exposed to whatever a prosecutor later says he was thinking.
That is why every rule in the table above is procedural rather than substantive. None of them tells a government what to decide. All of them require it to say why. The whole of this part is a description of what happens when very large decisions are made and that second step is skipped — and it is skipped, on the evidence of sixteen parts, by governments of every party, because nothing compels it.
9.4 — The question turned on this book
This series assumes that its own method survives its hardest case. Part One built a six-rung evidence scale and named the spectrum allocation as the worked example — the case where a real audit finding and a real acquittal both had to go on the page. Sixteen parts later, here it is.
So test the method against its own example. Chapter Eight found that in all four cases the allocation question went against the government and the criminal question did not stick. That is exactly what a graded account is built to show, and no other framework in circulation shows it: the political defence of this government says the acquittals settle everything, and its political prosecution says the audit figures do.
But now notice what the method cost. Sixteen parts of careful grading have produced, on the largest scandal in modern Indian history, a conclusion that reads as: something improper certainly happened, we cannot say anybody committed a crime, and we cannot say why the prosecutions failed.
That is honest. It is also, as advocacy, close to useless — and this book is an advocacy document that says so on its cover.
The uncomfortable finding is that the method and the purpose are in tension, and this is where the tension is greatest. A prosecutor who grades his evidence properly will sometimes have to tell the court that his best case cannot be proved. Part One chose that trade deliberately. Chapter Ten records what it has cost.
Naming that cost openly, in the middle of the part rather than in a footnote, is the only honest thing to do with it.
The defence is real: every criminal charge failed, the coal method dates from 1993, the building record is large, and the scandals are visible because of a transparency law this government passed.
It loses on four: six years between deciding to auction coal and doing it; the changed cut-off date in January 2008; a Lokpal legislated and not constituted for five years; and a communal violence law never passed in ten.
And the deepest problem is this book’s. Sixteen parts of careful grading, applied to the biggest scandal of the era, produce a conclusion too honest to be useful as advocacy.
10An Honest List of What We Do Not Know
Every part ends here. This one has the largest single item in the series on its list.
10.1 — Not established
Whether any minister of this government took money in any of these four matters. Not established. No court has found it. This book does not assert it and has not asserted it anywhere in this part.
Why the prosecutions failed. Chapter Eight set out the three available explanations — no crime, crimes badly prosecuted, or cases brought for other reasons — and said this book cannot choose between them. That remains the position. It is the largest unresolved question in nineteen parts, because the answer determines whether this part is about a scandal or about the manufacture of one.
What the true value of the allocations was. The audit figures are presumptive and their range is enormous. The stake sales described in Chapter Four are the only hard market evidence and they cover a handful of companies, not the whole allocation.
Whether a joint parliamentary committee would have established anything. One was demanded through an entire lost session in 2010 and eventually constituted. Its report was disputed by its own members. Part Fourteen found the same shape twice and Part Fifteen a third time.
10.2 — Where this part may be wrong
The status of the spectrum appeal. Leave to appeal was granted by the Delhi High Court in March 2024 and the substantive appeal is pending. If it succeeds, Chapter Eight’s right-hand column changes materially and this part will read very differently. A reader in five years should check it before relying on Table 5.
The coal cases are still concluding. The most recent judgment recorded here is from March 2026. Others may follow.
The Adarsh building. The Bombay High Court ordered it demolished in April 2016. This book has not been able to establish what has happened to that order since, and says so rather than guessing. Nothing in Chapter Six depends on it — the commission’s findings stand whether the building stands or not — but a reader who wants the end of the story should not take this part’s silence for an answer.
The audit figures. This part uses the CAG’s published numbers and the CBI’s charge sheet number and states what each measures. Reasonable people dispute all of them, including a convicted Coal Secretary whose technical criticism Chapter Five sets out at length.
The placement of Chapter Three. Explaining presumptive loss before describing any case is a choice, and a hostile reader may say the frame was built to soften what follows. It was built because Chapter Eight is unreadable without it, but that reading is available and cannot be disproved.
10.3 — What this part does to the rest of the series
Three findings here reach backwards and should be recorded, because Part Nineteen has to consolidate them.
The map appears for the third time. Part Five found the mineral belt — Jharkhand, Odisha, Chhattisgarh, Madhya Pradesh, Bengal — losing its only advantage to freight equalisation from 1952. Part Thirteen found the same belt at the overlap of the scheduled areas, the minerals and the insurgency. Part Sixteen finds it as the place the coal blocks were. Three parts, three different arguments, one map, and the same states are on the wrong end of all three.
The sanction requirement appears for the third time. Part Thirteen, Part Fifteen, and Chapter Six here. In each case a permission requirement described as a safeguard turned out to relocate the decision to somebody political. The rule that would fix it is in three separate tables now and has never been enacted.
The instrument was always older than the scandal. Part Nine found every Emergency instrument existing before June 1975. Part Ten found not one Punjab instrument that was a Punjab instrument. Part Fourteen found the terms-of-reference device in 1963 and 1992. Chapter Five here found the coal method dating from 1993 and struck down across five governments. This is now the most repeated finding in the series, and it is the one that most damages the framing of any single part as a scandal about a particular government.
It also, and this is the part a prosecution document has to say out loud, strengthens the charge against the party as a continuous actor and weakens almost every charge against any individual government. Part One’s founding box said that treating Congress as one thing across seventy-nine years cuts both ways. Sixteen parts in, it is now possible to say which way it cuts harder, and the answer is: against the party, and much less against the people who happened to be in office.
10.4 — What would settle it
Two things, and neither is a secret document.
The first is the spectrum appeal. It has been pending since 2018. Six years passed before it was admitted. A High Court judgment on the merits would establish whether the 2017 acquittals were sound, and it is the only instrument that can. Until then, the largest question in this part is open because a court has not reached it, not because anybody is hiding anything.
The second is an audit of the audits. Nobody has ever gone back over the presumptive-loss estimates of 2010 and 2012 with the actual outcomes now available — the auction prices realised after 2012 and 2015, the value of the cancelled coal blocks when reallocated, the stake sales. That comparison would show how good the estimates were.
Part Five ended by naming the calculation that would settle freight equalisation, which nobody has done. Part Fourteen named a second and Part Fifteen a third. This is the fourth, and like the others it is arithmetic. The data has existed for a decade.
One closing observation, and it is the most uncomfortable in this part.
The government that lost office in 2014 over these allegations had, in 2005, passed the law that made the file notings public. Chapter Nine listed that among its defences. It is also the reason there is a Part Sixteen at all.
Sixteen parts of this series have ended by naming something the state was holding back. Part Six named a report withheld since 1963; Part Fourteen, nine days of file notes from 1992; Part Fifteen, ninety pages from 2009. This part has the opposite problem and it is not a smaller one. Everything is public, the audit reports are on the internet, the judgments are published — and the central question is still unanswered, after sixteen years, because no court has reached it and no one has done the arithmetic.
Part Thirteen found an unmade addition where every component existed and nothing was concealed. This is that finding at national scale: total disclosure, and no conclusion.
Nobody has established that any minister of this government took money. This book does not say it.
Nobody has established why the prosecutions failed either. That is the largest open question in nineteen parts, because the answer decides whether this part is about a scandal or the manufacture of one.
And nothing is being hidden. The documents are all public. The appeal has been pending since 2018. The arithmetic has never been done.
Sources & further reading — Part 16
The State Ledger
Part Five committed every subsequent part to carrying this table: which states were affected by the material in this part, and how. It accumulates across the series and consolidates in Part Nineteen.
| State | What happened, 2009–2014 | Standing |
|---|---|---|
| Delhi | The Commonwealth Games and their procurement. Nineteen FIRs; closure reports accepted 2016 and 2025. The spectrum trial and appeal. | Charge; nothing established |
| Maharashtra | Adarsh. A Chief Minister resigned; a judicial commission found a quarter of the members ineligible; sanction to prosecute granted, then quashed. | Commission-found; no conviction |
| Jharkhand · Chhattisgarh · Odisha · Madhya Pradesh · West Bengal · Maharashtra | The coal belt. Most of the 214 cancelled blocks lie in these states — the same mineral belt Part Five identified as the loser under freight equalisation and Part Thirteen as the overlap of the three maps. | Court-found; the third time these states appear |
| Tamil Nadu | The spectrum allocation was made by a minister of a Tamil Nadu coalition partner; the money-laundering case concerned a Chennai television channel. All acquitted. | Charge; acquitted |
| Every state | Education as a right (2009); food security (2013); the 1894 land acquisition Act replaced (2013). | The credit column |
The third row is worth pausing on. Part Five found the mineral states losing their only advantage to freight equalisation from 1952. Part Thirteen found the same belt at the overlap of the scheduled areas, the minerals and the insurgency. Part Sixteen finds it again, as the place the coal blocks were. Three parts, three different arguments, one map.
Timeline
| Date | Event |
|---|---|
| 1993 | Coal block allocation by screening committee begins. Part Fourteen’s government. |
| 10 Jan 2008 | 122 telecom licences issued at 2001 prices, first-come-first-served, after the cut-off date is changed. |
| May 2009 | Congress wins 206 seats; UPA 262. No outside support needed. |
| Aug 2009 | Right to Education Act passed. |
| Oct 2010 | Commonwealth Games held in Delhi. |
| Nov 2010 | CAG report on spectrum: the ₹1,76,645 crore figure. Ashok Chavan resigns over Adarsh. |
| Winter 2010 | A parliamentary session is lost almost entirely to the demand for a joint committee on spectrum. |
| Jan–Apr 2011 | Kalmadi removed, then arrested. Anna Hazare’s first fast. |
| Feb 2011 | A. Raja arrested. He is held about fifteen months. |
| 2 Feb 2012 | Supreme Court cancels all 122 licences; holds the method arbitrary. |
| Mar 2012 | CAG draft on coal leaks: ₹10.67 lakh crore. |
| Aug 2012 | CAG final report on coal tabled: ₹1.86 lakh crore. |
| Nov 2012 | First post-cancellation spectrum auction. Largely unsold; the Finance Minister calls the ₹1.76 lakh crore figure “a pure myth”. |
| Dec 2012 | The Delhi gang rape. Justice Verma Committee appointed. |
| 2013 | The Patil Commission reports on Adarsh. Criminal Law (Amendment) Act. Food Security Act. Land Acquisition Act. |
| Jul 2013 | Lily Thomas: convicted legislators disqualified immediately. |
| 27 Sep 2013 | The ordinance reversing it is publicly torn up at a press conference. It is withdrawn. |
| Dec 2013 | Lokpal and Lokayuktas Act passed. |
| Jan 2014 | CBI files a closure report on the main Commonwealth Games contracts. |
| 25 Aug 2014 | Supreme Court: coal allocations since 1993 arbitrary and illegal. |
| 24 Sep 2014 | 214 of 218 coal blocks cancelled; ₹295 per tonne levy imposed. |
| May 2014 | Congress reduced to 44 seats. UPA 59. |
| Feb 2015 | Coal blocks re-auctioned. The government says proceeds will exceed the CAG estimate — a projection over thirty years. |
| Mar 2015 | Spectrum auction raises about ₹1.10 lakh crore. |
| 2016 | First coal convictions. Bombay High Court orders Adarsh demolished. Games closure report accepted. |
| 2017 | H. C. Gupta and two coal ministry officials convicted. Adarsh prosecution sanction quashed in December. |
| 21 Dec 2017 | All accused acquitted in all three spectrum cases. |
| Mar 2018 | CBI and Enforcement Directorate appeal. |
| Mar 2019 | The first Lokpal is appointed — five years and three months after the Act. |
| Mar 2024 | Delhi High Court admits the spectrum appeal, six years and about thirty hearings later. |
| Apr 2025 | Enforcement Directorate closure report in the Games case accepted. |
| Mar 2026 | Acquittals in the Bander coal block case close the oldest pending coal matter. |
Words Used in This Part
| CAG | Comptroller and Auditor General. The constitutional auditor. Reports to Parliament; cannot charge or convict anybody. |
| Captive block | A coal deposit given to a company to mine for its own plant rather than to sell. |
| Charge sheet | The document by which an agency formally accuses named people and takes a case to court. |
| Closure report | A filing by an investigating agency asking a court to end a case because the evidence does not support it. The court may accept or reject it. |
| Discharge | A court’s ruling that there is not enough material even to begin a trial. |
| First-come-first-served | Allocating a scarce public asset by order of application rather than by price or merit. |
| Leave to appeal | Permission from a higher court to bring an appeal. A threshold step, not the appeal itself. In the spectrum case it took six years. |
| Lokpal | A national anti-corruption ombudsman. Legislated December 2013; first appointment March 2019. |
| Ordinance | A law made by the executive when Parliament is not sitting. It lapses unless Parliament passes it. |
| Presumptive loss | The gap between what a government received and what an auditor calculates it could have received under a method it did not use. Chapter Three. |
| Prosecution sanction | Permission required before certain public servants can be prosecuted. See also Part Thirteen and Part Fifteen. |
| Public trust doctrine | The principle that natural resources are held by the state for the people and must be allocated for the common good. The basis of the 2012 and 2014 judgments. |
| Screening committee | Officials deciding who gets a coal block, with no bidding. In use from 1993 to 2010. |
| Spectrum | Radio frequencies used for mobile communication. Finite, owned by the state. |
| Terms of reference | The written instructions creating an inquiry and limiting what it may examine. See Part Fourteen, Chapter Five. |
A Word on Sources
Everything dated in this part was checked against a source rather than recalled, and where sources disagree the disagreement is on the page.
This part rests on public documents to an unusual degree. The CAG reports on spectrum, coal and the Commonwealth Games were laid before Parliament. The Supreme Court judgments of 2 February 2012 and 25 August and 24 September 2014 are reported. The special court’s acquittal of 21 December 2017, the coal convictions from 2016 onwards, the Bombay High Court’s order of December 2017 quashing the Adarsh sanction, and the closure reports accepted in 2016 and 2025 are all court records.
Where figures differ, both are given. The presumptive-loss estimates are shown in Table 1 with the comparison each rests on. The number of coal blocks cancelled is given as 214, which is the figure in most reports of the judgment; some sources say 204, and a reader who meets that figure should know the discrepancy exists.
Nothing in this part is graded above what a court or a commission actually found. The single most important consequence of that rule is negative: there is no claim anywhere in these ten chapters that a named minister took money, because no such finding exists.
Two things are live and may have moved. The spectrum appeal admitted in March 2024 is pending, and Chapter Ten says what happens to Table 5 if it succeeds. Coal cases were still being decided as recently as March 2026.
A note from the author
I have been building towards this part since Part One, and I want to record what writing it did to my own view, because the change was larger than I expected.
I started this series believing that the UPA II scandals were the clearest material I would have. Everybody knows the numbers. The government fell over them. It seemed like the part that would write itself.
What I found is that the numbers are softer than I understood, that the criminal cases produced almost nothing, and that the findings which do survive are narrower and duller than the ones in circulation — the six years on coal, a moved cut-off date, a Lokpal not appointed. Those are real and I have made them the charge. They are not what anybody argues about.
The place where this could have pulled the writing is Chapter Three. I could have written a much better polemic by putting the presumptive-loss explanation in Chapter Ten, as a caveat at the end, after nine chapters of scandal. That is where a caveat goes when you want it read last and remembered least. Putting it third, before any of the cases, cost this part most of its force, and I did it because Part One committed me to it in front of a reader who had not yet read anything.
I also want to be honest about Chapter Nine’s last box, because it is the closest this series comes to arguing against itself. Sixteen parts of grading evidence carefully have produced, on the biggest case, a verdict that is mostly a description of what is not known. If I had written this part first I might have chosen a different method. Having chosen it, I am not going to abandon it at the one point where abandoning it would be popular — but I am not going to pretend the cost is small either.
One last thing, which is about the next part rather than this one. Part Seventeen collects the statements — the things Congress leaders actually said, with dates and venues. After sixteen parts of arguing about what documents establish, it will be a relief to work with material where the only question is whether the words were spoken. I mention it because I noticed myself looking forward to it, and a writer who is relieved to leave a subject should say so where the reader can see it.
Lovepreet Singh
Part Sixteen of The Congress Record
The Living Archive · misterlove.in
Part Seventeen is What Was Said: the statements, each with date, venue and full wording. The comparison with the government of 2014 onwards is Part Eighteen, measured on the same ruler, and the second series auditing that government on this scale is still promised.