AT LONG LAST, THE ASPIRANT TO HIGH OFFICE HAS triumphed. Whether through inheritance, coup, election, revolt, murder, or mayhem, he has seized power. Now he faces a new challenge: hanging on to it.

As Sergeant Doe’s brutal career has taught us, rising to a high position often requires skills altogether different from those needed to maintain control. And even the rules for surviving in power do not always resemble the skills necessary for ruling well. The novelist Italo Calvino has clearly and succinctly described the tribulations of those who have risen to power: “The throne, once you have been crowned, is where you had best remain seated, without moving, day and night. All your previous life has been only a waiting to become king; now you are king; you have only to reign. And what is reigning if not this long wait? Waiting for the moment when you will be deposed, when you will have to take leave of the throne, the scepter, the crown, and your head.”¹

What, then, must a newly minted leader do to keep his (or her) head? A good starting place is to shore up the coalition of supporters. This may seem like a simple enough task. After all, as we’ve seen, the heights of power are unattainable without the backing of a coalition strong enough to beat back rivals. However, a wise leader does not count too much on those who helped her gain power. Remember the fate of many of Fidel Castro’s closest allies. After toppling the previous leader, it’s only a matter of time until they realize that they can do the same again.

A prudent new incumbent will act swiftly to get some of them out of the way and bring in others whose interests more strongly assure their future loyalty. Only after sacking, shuffling, and shrinking their particular set of essentials can a leader’s future tenure be assured.

Nor is this only true of dictators. To see this urge to build a modified coalition at work in the seemingly less ferocious world of business, let’s take a look at Carly Fiorina’s rise and fall as CEO at Hewlett-Packard.

Governance in Pursuit of Heads

CEOs, just like national leaders, are susceptible to removal. Being vulnerable to a coup, they need to modify the corporate coalition (usually the board of directors and senior management), bringing in loyalists and getting rid of potential troublemakers. Usually, they have a large potential pool of people to draw from and prior experience to help guide their choices. But, also like national leaders, they face resistance from some members of their inherited coalition and that may be hard to overcome.

Most publicly traded corporations have millions of interchangeables (their shareholders), a considerably smaller set of influentials (big individual shareholders and institutional shareholders), and a small group of essentials, often not more than ten to fifteen people. In a group of this size, even seemingly minor variations in the number of coalition members can have profound consequences for how a company is run. As we will see, this was particularly true for Hewlett-Packard (HP), because, as in all companies, small shifts in coalition numbers can lead to large percentage changes in the expected mix of corporate rewards.

In the case of HP, the CEO’s winning coalition made up a relatively large fraction of the real selectorate because ownership is heavily concentrated in a few hands. That is, we might count corporate coalition size in terms of the number of its members or in terms of the number of shares owned by them. In HP’s case, the essential bloc and the influential bloc are a tiny part of the total selectorate because the families of the company’s founders, William Hewlett and David Packard, retain significant ownership just as was true of the Ford Motor Company, Hallmark Cards, and quite a few other businesses for many years.

Involvement in a corporation can yield benefits, just like any other form of government. These benefits can take the form of rewards given to everyone or private payments directed just to the essentials. In a corporate setting, private benefits typically come as personal compensation in the form of salary, perks, and stock options. Rewards to everyone—what economists call “public goods”—take the shape of dividends (an equal amount per share) and increased stock value. When the winning coalition is sufficiently large that private rewards are an inefficient mechanism for the CEO to buy the loyalty of essentials, public goods tend to be the benefit of choice. Usually, coalition members are eager to receive private benefits. However, dividends and growth in share value are preferred over private rewards by very large shareholders who happen also to be in the winning coalition—this would make them the biggest recipients of the rewards that go to all shareholders. That was precisely the situation in HP, where the Hewlett and Packard families owned a substantial percentage of the company.

Who makes up the essentials in a corporation? The coalition typically includes no more than a few people in senior management and the members of the board of directors. These directors are drawn from a mix of senior management in the company, large institutional shareholders, handpicked friends and relatives of the CEO (generally described as civic leaders, no doubt), and the CEO herself. In the parlance of economists who study corporations, the makeup of these boards boils down to insiders (employees), grey members (friends, relatives), and outsiders. One part of any corporate board’s duties is to appoint, retain, or remove CEOs. Generally CEOs keep their job for a long time and that certainly was true of HP’s first CEO, founder David Packard. He was replaced in 1992 by an insider, Lewis Platt, who had worked for the firm since the 1960s. Platt retired in 1999 and was replaced by outsider Carly Fiorina. The HP board has repeatedly deposed CEOs since then.

It should be obvious that any board members involved in deposing the former CEO have the potential to be a problem for a new CEO. Having once been a coup maker, there is little reason to doubt that they stand ready to start trouble once again if they think the circumstances warrant it. And what could those circumstances be but application of one or more of the rules of governance we set out earlier, especially if that application harms their interests.

Research into CEO longevity teaches us, not surprisingly, that time in office lengthens as one maintains close personal ties to members of the board. Just as sons and daughters may make attractive inheritors of the mantle of power in a dictatorship, friends, relatives, and fellow employees can generate the expectation of more loyal supporters after power is achieved. This logic probably contributed to Lewis Platt’s elevation to CEO of HP. Putting more outsiders on a board translates on average into better returns for shareholders, a benefit to everyone. At the same time, it also translates into greater risk for the CEO.² Since the CEO’s interest is rarely the same as the shareholder’s interest, CEOs prefer to avoid outsider board members if they can.

Corporate problems, especially those serious enough to oust a sitting CEO, can serve to galvanize attention and enhance oversight by the board, making existing coalitions less reliable. Furthermore it is likely that the new, replacement CEO will face real impediments to his efforts to create and shape a board of directors in the wake of an older CEO’s deposition. After all, the old board members did not get rid of the prior incumbent with the idea that they would also make it easy for the successor CEO to get rid of them. Nevertheless, any new CEO worth her salt will try to do just that. The long-lasting CEOs are the ones who succeed.

Carly Fiorina became Hewlett-Packard’s CEO in 1999. After six turbulent years she was deposed from that position and as chairwoman of the board in early 2005. Prior to being removed she was the target of an unsuccessful proxy fight mounted by Walter Hewlett and David Woodley Packard, sons of HP’s founders. The board, in keeping with the power of inherited insider influence, also included Susan Orr, founder David Packard’s daughter. All were individuals with big financial stakes in HP. Furthermore, as big shareholders Hewlett, Packard, and Orr were more concerned about HP’s overall performance than about any private benefits they got from being on the board. Good news for shareholders—potentially bad news for Fiorina.

The board that selected Fiorina as CEO consisted of fourteen members. As we’ve seen, three were relatives of HP’s founders; three more were current or retired HP employees.³ Fiorina’s initial board, in other words, had a substantial group of insider and grey members who were not of her choosing and who had big stakes in the corporation’s stock value. It is not hard to see that Carly Fiorina needed to make changes to build a leaner board with stronger attachments to her. It would not be easy—while the previous board selected her, they were not her handpicked loyalists.

She achieved results, nonetheless. A year after Fiorina’s ascension, HP’s proxy statement to its shareholders in 2000 listed only eleven board members, 20 percent fewer than the group that selected her. Three, including David Woodley Packard, were gone. As Fiorina became more entrenched in her position, the board continued to shrink—the 2001 statement listed only ten board members, a reduction of nearly 30 percent from the board she originally inherited. Seemingly growing more secure in her control, Fiorina launched an effort to merge Compaq with HP, an effort with both beneficial prospects and serious risks for her continued rule.

Naturally, Fiorina presented the merger as a boon for HP and its shareholders. As Fiorina explained on February 4, 2002,

It is a rare opportunity when a technology company can advance its market position substantially and reduce its cost structure substantially at the same time. And this is possible because Compaq and HP are in the same businesses, pursuing the same strategies, in the same markets, with complementary capabilities. So, yes, we thought about a go-slow approach. But, we concluded, after two-and-a-half years of careful deliberation and preparation, that standing still had enormous risks.... Standing still means choosing the path of retreat, not leadership.⁴

There is no reason to doubt the sincerity of Fiorina’s expectations for the Compaq merger. But it is instructive to examine a major indicator of how Fiorina’s appointment and how her views meshed with broader market sentiments. The day before the announcement of Carly Fiorina’s appointment as HP’s new CEO, HP’s shares traded at $53.43. The market’s reaction to her appointment can reasonably be described as uncertain. The price of HP shares was flat immediately following the announcement and then began a decline, falling to under $39 by mid-October 1999, about three months later. Of course, markets are forward looking and so investors were watching and learning, modifying their expectations as Fiorina took charge. The news and modified expectations must have been good for a while because by early April 2000, HP’s shares had risen markedly to about $78. But good feelings and good circumstances were not to prevail for long. After April 7 the share price went into a tailspin, bottoming out in September 2002 at around $12 a share, and significantly underperforming the major stock market indexes. By the time Fiorina resigned in February 2005, HP’s share price had only rebounded to about $20.

With respect to the Compaq merger, the market was similarly pessimistic. The plan to merge with Compaq was announced on September 3, 2001. The shares rose on the news, with a peak in December of that year of about $23, though still well below the value the day before Carly Fiorina became CEO. Over the period from July 1999 (the announcement of Fiorina’s appointment) to the end of December 2001, the adjusted Dow Jones index fell 9.4 percent while HP’s adjusted share price fell 47 percent.

From the perspective of any big investor in HP, including the Hewlett and Packard families, Fiorina must have looked like a disaster. Their company was doing worse than the general stock market; their fortunes were being hammered. She was a CEO in trouble. Nevertheless, the upward tick in the share value indicated a renewed, if temporary, boost of optimism at the announced intention to merge with Compaq. But markets don’t like infighting, and when Walter Hewlett and David Woodley Packard declared their opposition to the merger, the gains were reversed. Soon the price collapsed even further, halving as it became apparent that there was to be a proxy fight in which Hewlett and Packard sought to muster support from enough shareholders to defeat the board’s proposed slate at the corporation’s annual meeting. No doubt Fiorina realized that she was going to be in for a tough time, perhaps even before her public announcement of the intended and eventually successfully completed merger. It also seems likely that she would already have known Hewlett and Packard’s views. We can only conclude that this was an intentional gamble on a major policy shift, one that could—and did—adversely affect the wealth of HP’s large shareholders (such as present or former board members Hewlett and Packard).

Looking at the Compaq-HP merger politically, we can see several critical themes emerging. Fiorina was already in some trouble because of declining share value. She had successfully diminished the board’s size and shuffled its membership, both wise choices for a CEO seeking longevity in office. Yet despite these actions, she still faced significant opposition from the inner circle of essentials and influentials. She had not yet secured the board’s loyalty. The Compaq merger might have made good business sense and could therefore have been good for the stock price, thus softening internal opposition to her. Or else, seeing the merger as a fait accompli, her opponents might have given up their fight. That didn’t happen. And the disgruntled board members, heavily invested as they were in HP’s stock value, could not be mollified with private rewards.

However, what in retrospect may seem like a political nonstarter at the time held great political advantages. What, for instance, had to be an implication for board composition of Fiorina’s multibillion dollar merger with Compaq? Once the deal was sealed Fiorina would have to bring some Compaq leaders onto the postmerger HP board. This could be done either by expanding the existing board to accommodate Compaq influentials or by pruning the existing board to make room for the new, Compaq representatives drawn from Compaq’s selectorate. Fiorina apparently saw that the merger would provide an opportunity to reconstitute the board, providing an undeniable opportunity to weaken the board faction that opposed her. That seems to be exactly what she tried to do.

Of course, her rivals would not sit idly by and be purged. Unless such a purge can be accomplished in the dark, presented as a fait accompli to the old group of influentials, the risk of failure is real. As it happens, Securities and Exchange Commission (SEC) regulations require disclosures, which make turning a board purge into a fait accompli extremely difficult when the opportunity to purge the board depends on a prospective merger.

There are two potential responses to a rebellion such as the one Fiorina faced over HP’s weak share price and the Compaq merger. A CEO can either purge essentials and boost the private benefits to remaining coalition members, or expand the coalition and increase rewards to the general selectorate of interchangeables (that is, shareholders). Having survived the proxy fight in 2002, Fiorina faced an eleven-member board that included five new members carried over from Compaq as part of the merger. HP’s board had shifted materially with only six previous HP board members on it. Since Fiorina had been the mover and shaker behind the Compaq deal it is reasonable to believe that she assumed the new members would be likely to work with her as opposed to lining up with board members who had supported Walter Hewlett’s fight against the merger. Walter Hewlett and Robert P. Wayman, meanwhile, left the board. By this time Fiorina had expanded the total board size by only one, from ten to eleven, while overseeing the departure of several old board members at the same time to make room for five Compaq representatives. Surely she had reason to believe she now enjoyed the support of a majority of the new board.

Perhaps in an effort to shore up the support of remaining old hands on the board, or perhaps coincidentally, there also was a notable shift in board compensation. Just before Fiorina became HP’s head, board members earned compensation (that is, private benefits) that ranged from $105,700 to $110,700. With Fiorina in office and the board diminished in size, this amount dropped slightly to $100,000–$105,000 and remained there in the years 2000–2003. But in 2004, according to HP’s 2005 proxy statement, board members received $200,000 to $220,000. During the same period, dividends remained steady at $0.32 a share annually and HP’s shares significantly underperformed the main stock market indexes. Clearly something was up: HP’s stock price performance was poor; dividends were steady; and directors’ pay doubled.

Fiorina’s board shuffling and their improved compensation seem aimed at getting the right loyalists in place to help her survive. Although the Compaq merger resulted in the board growing from ten to eleven, what is most noteworthy is that this net growth of one member was achieved while adding five new members (one of whom stepped down at the end of the year). So, the old members constituted only about half the board, shifting the potential balance of power toward Fiorina. Presumably that is just what she hoped, although it is not how things turned out.

Expanding the board was not, and generally is not, the optimal response to a threat from within. To her credit, in terms of political logic, she significantly expanded the size of the interchangeables by adding Compaq’s shareholders to HP’s list of shareholders. This normally helps to induce strengthened loyalty, but declining share value could not have been good for new HP board members, who had been heavily invested in Compaq since their economic well-being was now tied to HP’s share performance. Nor could Fiorina mollify HP’s large shareholders on the board with better board compensation, since their welfare depended on producing the “public good” of greater returns to shareholders. Those grey board members who owned lots of shares made the seemingly small board of eleven actually pretty large in terms of shares they could vote.

Under enormous pressure, Carly Fiorina stepped down. She was replaced by Patricia Dunn as chairwoman, with HP’s Chief Financial Officer (CFO), Robert Wayman, emerging again as a significant HP player. He was made interim CEO. Wayman, unable or uninterested in translating his interim position into a full-time job, stepped down a month later while continuing in his role as a member of the board and an HP employee. Mark Hurd in turn replaced him as CEO.

In the immediate aftermath of Fiorina’s ouster the board separated two key positions, CEO and chairperson, presumably in a good Montesquieu-like effort to promote the separation of powers and protect themselves against future adverse choices by the CEO. If that was their intention, they certainly failed. Following Hurd’s ascent to the position of CEO he successfully brought the two posts back under one person’s control: his own.

Within a year of Fiorina’s ouster, all the leading coup makers who acted against her were gone. Mark Hurd had risen to the top and, as suggested by the quote from Italo Calvino, he had to watch day and night to keep his head. Four years later, despite stellar HP performance, Hurd was, in turn, forced out amidst a personal scandal. This is the essential lesson of politics: in the end ruling is the objective, not ruling well.

The Perils of Meritocracy

One lesson to be learned from Mark Hurd’s ultimate removal at HP is that doing a good job is not enough to ensure political survival. That is true whether one is running a business, a charity, or a national government. How much a leader’s performance influences remaining in office is a highly subjective matter. It might seem obvious that it is important to have people in the coalition of key backers who are competent at performing the duties associated with implementing the leader’s policies. But autocracy isn’t about good governance. It’s about what’s good for the leader, not what’s good for the people. In fact, having competent ministers, or competent corporate board members, can be a dangerous mistake. Competent people, after all, are potential (and potentially competent) rivals.

The three most important characteristics of a coalition are: (1) Loyalty; (2) Loyalty; (3) Loyalty. Successful leaders surround themselves with trusted friends and family, and rid themselves of any ambitious supporters. Carly Fiorina had a hard time achieving that objective and as a result she failed to last long. Fidel Castro, by contrast, was a master (of course, he had fewer impediments to overcome in what he could do than did Fiorina) and he lasted in power for nearly half a century.

The implications of this aspect of political logic are profound, particularly in small coalition governments. Saddam Hussein in Iraq, like Idi Amin in Uganda and so many other eventual national leaders, started as a street thug. Autocrats don’t need West Point graduates to protect them. Once in power, people like Amin and Hussein wisely surround themselves with trusted members of their own tribe or clan, installing them in the most important positions—those involving force and money—and killing anyone that may turn out to be a rival.

Saddam Hussein came to power after compelling his predecessor (and cousin) Ahmed Hassan al-Bakr to resign in 1979.⁵ Before that, however, he had carefully laid the groundwork for his control over Iraq. In 1972, for instance, he spearheaded the nationalization of international oil interests in Iraq. Oil, of course, was and is where the money is in Iraq, so he had fulfilled the essential ingredient to come to power: he knew where the money was. Once in power, he ruthlessly pruned his support base.

Just six days after President al-Bakr “resigned,” Saddam Hussein convened a national assembly of the ruling Ba’ath Party’s leaders (the Revolutionary Command Council). The assembly was videotaped at Saddam Hussein’s insistence. During the session, Muhyi Abdel-Hussein, secretary of the Revolutionary Command Council, read out a confession that he plotted against Saddam Hussein, and then sixty-eight more “enemies of the state” were named as coconspirators. Each, one at a time, was removed from the assembly. Twenty-two were sentenced to death by firing squad and summarily executed by members of the Ba’ath Party, each branch of which was required to send a delegate with a rifle to participate in the executions. Hundreds more were executed within the next few days. Saddam Hussein’s biographer asked Saddam about the decision to eliminate these people, most of whom had risen in the ranks of the Ba’ath Party with Saddam’s support. He reports, “The answer was that as long as there is a revolution, there will be a counter-revolution.”⁶ As we said before, those who can bring a leader to power can also bring the leader down. It is best to shrink the ranks of those who represent a threat and keep those who are most trusted to be loyal.

How competent were the approximately 450 Ba’ath leaders who were executed as part of Saddam’s consolidation of power? It is difficult to say from this remove, but we do know that among their ranks were professors, military officers, lawyers, judges, business leaders, journalists, religious leaders, and many other well-educated and accomplished men. For good measure, Hussein also threw in leaders of competing political parties who, after all, might have conspired to replace him.

Survivors included people like Saddam’s cousin, “Chemical Ali,” Ali Hassan al-Majid. Chemical Ali most notably demonstrated his loyalty in 1988 when, under orders from Saddam, he launched a successful campaign to commit genocide against Iraq’s restive Kurds. Long before that al-Majid had established his commitment to Saddam Hussein. In the infamous videotape mentioned earlier, al-Majid is seen speaking to Saddam, saying, “What you have done in the past was good. What you will do in the future is good. But there’s one small point. You have been too gentle, too merciful.”⁷ Unlike many who were executed following the July 22, 1979, party assembly, al-Majid, previously a motorcycle courier/delivery boy, had little formal education. Although he held the posts of defense minister, interior minister, and head of Iraq’s intelligence service, it seems his main area of competence was murder.

Saddam Hussein’s pattern of appointments is quite typical. His successor, Prime Minister Nouri al-Maliki, purged the security services of all Sunnis and replaced them with Shia supporters, albeit with a gentler hand than his predecessor.⁸ These replacements did not have the experience and training of the existing security personnel. Both leaders knew that it is better to have loyal incompetents than competent rivals.

Sometimes, of course, having competent advisers is unavoidable. Byzantine, Mughal, Chinese, Caliphate, and other emperors devised a creative solution that guaranteed that these advisers didn’t become rivals: They all relied on eunuchs at various times. In the Byzantine Empire in the ninth and tenth centuries, the three most senior posts below emperor were held almost exclusively by eunuchs. The most senior position of Grand Administrator had evolved from the position of Prefect of the Sacred Bedchamber and included the duties of posting eunuch guards and watching over the sleeping emperor. Michael III made an exception and gave this position to his favorite, Basil, rather than a eunuch. This decision cost him his life. When Basil perceived that Michael was starting to favor another courtier, he murdered the emperor and seized the throne.⁹

Even in modern times the principle of choosing close advisers who cannot rise to the top spot remains good advice. It is surely no coincidence that Saddam Hussein as president of Islamic Iraq had a Christian, Tariq Aziz, as his number two.

Keep Essentials Off-Balance

What we can begin to appreciate is that no matter how well a tyrant builds his coalitions, it is important to keep the coalition itself off-balance. Familiarity breeds contempt. As noted, the best way to stay in power is to keep the coalition small and, crucially, to make sure that everyone in it knows that there are plenty of replacements for them. This is why you will often read about regular elections in tyrannical states. Everyone knows that these elections don’t count, and yet people go along with them. Rigged elections are not about picking leaders. They are not about gaining legitimacy. How can an election be legitimate when its outcome is known before the vote even occurs? Rigged elections are a warning to powerful politicians that they are expendable if they deviate from the leader’s desired path.

Vladimir Ilyich Lenin was the first to really exploit the idea of substitute coalition members. In a one-party state, he nonetheless perfected a rigged election, universal adult suffrage system. Any action he took—say, sending so-and-so to Siberia—was the will of the people, and any of the people in the replacement pool had a chance, albeit a slight one, of being called up to serve as an influential or maybe even an essential somewhere down the line. Everybody in the Soviet selectorate could, with a very small probability, grow up to be general secretary of the Communist Party, just like the petty criminal Joseph Stalin and the uneducated Nikita Khrushchev. Those already in the inner circle knew they had to stay in line to keep their day jobs. Bravo, Lenin.

Although Lenin perfected the system and probably came up with it on his own, the always fascinating country of Liberia experimented earlier on with the same phenomenon. Prior to Samuel Doe’s takeover, Liberia had been ruled by the True Whig Party. The country originated when a number of American liberal organizations, appalled by the evils of slavery, paid to repatriate former slaves to West Africa. Despite the nation’s philanthropic origins, the most important lesson the former slaves took from their experiences appears to be that slavery and forced labor worked much better for the masters than the slaves. These former slaves instituted universal adult suffrage in 1904, but with a property qualification that effectively excluded indigenous Africans from becoming insiders, making the selectorate large but the influential group relatively small. Thus, they established a system run for a small group of insiders despite the appearance of a universal franchise. This structure provided for strong loyalty to the incumbent that ensured the opportunity to suppress any opposition that might arise to their forced labor policies, a system whose policies differed from Soviet ones but whose security in office was the same.¹⁰

Virtually every publicly traded company in the world has adopted the Leninist rigged-election system and for much the same reasons. It, along with a packed board, is one of the major factors ensuring that poorly performing CEOs hardly ever get fired. Carly Fiorina had the misfortune of heading a company that might have looked like a rigged election autocracy but up close and personal remained more akin to a monarchy. Although there were millions of shareholders who in theory could shape HP policy, so many shares were concentrated in a few hands that HP had more of the characteristics of a small coalition drawn from a small group of influentials within a mostly small, concentrated group of interchangeables; that is, members of the Hewlett and Packard families.

The essence of keeping coalition members off-balance is to make sure that their loyalty is paid for and that they know they will be ousted if their reliability is in doubt. The USSR’s Mikhail Gorbachev, thought to be a good guy in western political circles, certainly understood the necessity of rewarding loyalty and shucking off all those whose faithfulness was questionable. He replaced much of the politburo within his first two years in office, picking and choosing from the Communist Party (the real selectorate) those most loyal to him. It turns out, though, that Gorbachev was much less ruthless than contemporaries of the autocratic class. He forced adversaries, like Boris Yeltsin, out of the politburo to be sure. But, as Yeltsin surely realized, he would have been killed under Stalin. Equally, he and many others must have known that it was much better to cross swords with Gorbachev, an intellectual reformer, than with such contemporaries as Mobutu Sese Seko of Zaire or even Deng Xiaoping of China. Deng, after all, used ruthless force to end the prodemocracy uprising at Tiananmen Square in 1989. Gorbachev, as we will see, did not hesitate to use force outside of Russia, but he also did not go around killing his political rivals. His reward was a short time in power first because he left himself vulnerable to a coup by hard-line communists and then because he allowed Yeltsin to resurrect himself politically, defeat the coup, and make himself into Gorbachev’s replacement.

The execution of opponents is a longstanding practice among most autocrats. We should not fail to appreciate the moral significance of Gorbachev’s restraint. Adolf Hitler, Mao Zedong, Fidel Castro, Samuel Doe, and so many others showed no such restraint. They had their erstwhile backers murdered once they worked out who was most likely to be loyal and who was not. We see a nicer version of such behavior as a routine part of corporate changes when there’s a new CEO. Although the CEO is supposed to answer to the board, it is commonplace for boards to be reconstituted after a new CEO comes to power; the tail apparently wags the dog.

Being purged from the initial coalition is often fatal. Hitler became chancellor of Germany on January 30, 1933. During his rise to power he relied heavily on the Sturmabteilung, a paramilitary force also known by the abbreviation, SA, or by a description of their uniforms, the Brownshirts. Hitler perceived the SA’s leader, Ernst Rohm, as a threat. He built up an alternative paramilitary, the Schutzstaffel, or SS, and then, on what became known as the night of the long knives, he ordered the assassination of at least eighty-five and possibly many hundreds of people between June 30 and July 2, 1934. Thousands more were imprisoned. Despite Rohm’s long term and essential backing (Rohm had been with Hitler during his failed 1923 Munich Beer Hall Putsch), Hitler showed no sentimentality. He replaced him with men like SS leader Heinrich Himmler, whom he deemed more loyal.

Robert Mugabe is likewise a master at keeping his coalition off-balance. He was elected president of Zimbabwe in 1980 following a negotiated settlement to a long civil war. The struggle against the white-only rule of the previous Rhodesian regime was led by two factions that crystallized into political parties behind their respective leaders: Robert Mugabe’s ZANU (Zimbabwe African National Union) and Joshua Nkomo’s ZAPU (Zimbabwe African People’s Union). Initially, Mugabe preached reconciliation:

If yesterday I fought you as an enemy, today you have become a friend and ally with the same national interest, loyalty, rights and duties as myself. If yesterday you hated me, you cannot avoid the love that binds you to me and me to you. . . . Draw a line under the past.... The wrongs of the past must now stand forgiven and forgotten. If ever we look to the past, let us do so for the lesson the past has taught us, namely that oppression and racism are inequalities that must never find scope in our political and social system. It could never be a correct justification that because the whites oppressed us yesterday when they had power, the blacks must oppress them today because they have power. An evil remains an evil whether practiced by white against black or black against white.¹¹

A naïve observer might have thought that Mugabe planned to bring ZAPU elites into his winning coalition. That might have made sense at the outset, but once ZANU’s power was consolidated there would be no reason to keep ZAPU loyalists around. And once Mugabe’s power was consolidated, he’d have no need to keep some of his old friends from ZANU around either.

Mugabe also reached out to many in the white community, and particularly former leaders and administrators, to help him run the country. Many whites who had feared the transition, began to refer to him as “Good Old Bob.” Mugabe needed their support. He could not run the country without them and he needed to know where the money was. In this he was greatly assisted by the international community. They pledged $900 million during his first year. However, once he was ensconced in power, Mugabe’s attitude changed.

In 1981 he called for a one party state and began arresting whites, saying “we will kill those snakes among us, we will smash them completely.” Mugabe was even harsher towards his former comrades in arms. He forced Nkomo out of the cabinet and sent a North Korean trained paramilitary group, the Fifth Brigade, to terrorize Matabeleland, Nkomo’s regional stronghold. As one ZANU minister put it, “Nkomo and his guerillas are germs in the country’s wounds and they will have to be cleaned up with iodine. The patient will scream a bit.” The operation was called Gukurahundi—a Shona word that means, Wind that blows away the chaff before the spring rains. Many veterans from the fight against white rule resisted. In retaliation Matabeleland was effectively sealed off and 400,000 people faced starvation. As one of Mugabe’s henchmen, a brigade officer, stated, “First you will eat your chickens, then your goats, then your cattle, then your donkeys. Then you will eat your children and finally you will eat the dissidents.”¹²

Mugabe needed the assistance of ZAPU fighters to defeat white only rule. He needed the assistance of white farmers and administrators and the international community to find the money to solidify his control over the state. Only when he was entrenched in power did “Good Old Bob” show his true colors.

Democrats Aren’t Angels

As we all know, the victor writes history. Leaders should therefore never refrain from cheating if they can get away with it. Democrats may have to put up with real and meaningful elections in order to stay in power, but it shouldn’t be shocking to see that whenever they can, they’ll happily take a page out of Lenin’s book. There’s no election better than a rigged one, so long as you’re the one rigging it.

The list of tried and trusted means of cheating is long. Just as quickly as electoral rules are created to outlaw corrupt practices, politicians find other means. For instance, leaders can restrict who is eligible and registered to vote and who is not. In Malaysia, under a system known as Operation IC, immigration is controlled so as to create demographics favorable to the incumbent party. New York City’s infamous Democratic Party machine, Tammany Hall, acquired its Irish flavor by meeting and recruiting immigrants as they left the boat, promising citizenship and jobs for their vote.

When leaders can’t restrict who is eligible to vote or else are unable to buy enough votes, they can use intimidation and violence to restrict access to polling places. North Indian states, such as Bihar and Uttar Pradesh, experience “booth capture,” where party supporters capture the polling place and cast every eligible voter’s vote for their party.

Cheating does not stop once ballots are cast, of course. Leaders never hesitate to miscount or destroy ballots. Coming to office and staying in office are the most important things in politics. And candidates who aren’t willing to cheat are typically beaten by those who are. Since democracies typically work out myriad ways to make cheating difficult, politicians in power in democracies have innovated any number of perfectly legal means to ensure their electoral victories and their continued rule.

One counterintuitive strategy is for leaders to encourage additional competitors. This is why some states have so many political parties, even though only one really wins. The conventional wisdom about America’s two-party system tells us that fringe parties allow for a more vibrant and responsive government. But even in multiparty states, there are always leading parties—you have to ask yourself whether the leading parties would allow the fringe parties to exist if they weren’t somehow serving their interests.

Tanzania’s parliament and presidency are perennially controlled by the Chama Cha Mapinduzi party (CCM), even though as many as seventeen parties routinely compete in Tanzania’s free and fair elections. The CCM government actually provided campaign financing, as we would expect, in an opaque way, to small parties until quite recently, thereby encouraging them to compete and divide the opposition vote. This makes it easier for the relatively centrist CCM to win. Although the CCM wins a large percentage of the vote, all it needs to win is one more vote than the second largest party in half the parliamentary constituencies. That turns out to mean the CCM needs much less than 10 percent in most districts. The number of supporters a party needs affects the kinds of policies it pursues. In those constituencies in Tanzania where an opposition party generates lots of votes, the CCM needs to appeal to many voters and therefore generally provides better health care, education, and services. In constituencies where the CCM needs fewer votes, cash transfers, such as vouchers for subsidized fertilizer, are more common.¹³

Multiparty democracy provides a similar means for one or two parties to dominate governments in democracies from Botswana to Japan and Israel. There is more to representing the people than just allowing them to vote, even when the vote is done honestly.

Designated seats for underrepresented minorities is another means by which leaders reduce the number of people upon whom they are dependent. Such policies are advertised as empowering minorities, whether they are women, or members of a particular caste or religion. In reality they empower leaders. That a candidate is elected by a small subset of the population reduces the number of essentials required to retain power. At a very basic level, electoral victory in a two-party parliamentary system requires the support of half the people in half the districts; that is, in principle, 25 percent of the voters. Suppose 10 percent of the seats were reserved for election by one specific group that happens to be geographically concentrated (such as gay voters in the Castro in our earlier account of Harvey Milk’s election in San Francisco). To retain half the seats in parliament, the incumbent party need only retain 40 percent of the regular single member district seats, which is readily done with just over 22 percent of the vote. So by focusing on districts in which the privileged minority is prevalent, a party can reduce the number of votes it requires by 12 percent.

Delegated positions also make it easier to form a small coalition. Consider Tanzania’s Parliament, the Bunge. There are 232 directly elected seats, seventy-five seats reserved for women who are nominated by the parties in relation to the number of seats they capture in the election, and five seats nominated by the Zanzibar Assembly. (Zanzibar is a beautiful island off the mainland that united with mainland Tanganyika in 1964 to form Tanzania.) In addition, the president gets to nominate ten cabinet appointees and an attorney general to serve in parliament. This gives a total of 323 seats, of which the president needs 162 to control the Bunge. Given that he appoints eleven, and that the CCM is regionally based in Zanzibar, he already controls sixteen seats. If the CCM wins 111 elected seats, then he controls parliament. That is, 111 directly elected seats, 16 appointed seats, and 35 of the appointed women’s seats (75 seats x 111/232), which totals 162. The CCM needs substantially less than half the directly elected seats. And as we have already seen, by funding many opposition parties the CCM can win many seats with less than a 10 percent vote share. In practice the president controls nearly all the women’s appointments and he tends to appoint women who lack an independent base of support. Indeed, few women win direct election to Tanzania’s parliament.

While Tanzania has free and fair elections, the reality is that the incumbent CCM party can sustain itself in office with as little as 5 percent of the vote. Of course, in most districts they get much more support because politicians find inventive ways to incentivize voters. One of these ways is the creation of voting blocs.¹⁴

Bloc Voting

Bloc voting is a feature common in many fledgling democracies. It was also the norm under party machines in large US cities. For instance, under the influence of Tammany Hall, whole neighborhoods in New York City would turn up to vote Democratic. Many of India’s electoral districts have followed a pattern similar to the old Tammany Hall. That is, a small group of local notables or village patrons can deliver their community’s vote and extract great rewards for themselves in return.

During Bueno de Mesquita’s time doing field work in India in 1969–1970 he observed firsthand how the quest for power coupled with the influence of power blocs undermined any notion of the pursuit of political principles other than the principles, win, and get paid off.

Senior people in villages and towns, and indeed, up and down the levels of governance in India’s states, would pledge to a particular party the support of those they led. In return, they would receive benefits and privileges. By and large, all the “clients” of these “patrons” followed their patron’s lead and voted for the designated party. What is most fascinating is that the affiliations between voters and parties need not have had any ideological rhyme or reason. In Uttar Pradesh, India’s most populace state, for instance, the free-market, anticommunist Swatantra Party, the socially conservative and anticommunist Jana Sangh Party, and the Communist Party of India formed a coalition government with each other following India’s 1967 election. This was true despite the Swatantra Party’s leadership’s description of the Communist Party of India as “public enemy number 1.” What did these parties have in common? Only their desire to band together and beat the Congress Party so as to enjoy the benefits of power. This sort of odd bedfellows coalition-building strategy was long rampant throughout India.¹⁵

Perhaps the most egregious case of bald opportunism occurred in the state of Bihar. There ideologically disparate parties formed a government, relying heavily on currying favor with the Raja of Ramgarh. The raja, owner of much of the mining interests in Bihar, switched parties every few months, bringing coalition governments down—and up—with him. Each time he switched, he garnered greater private goods for himself and his backers, including the dismissal of criminal charges against him. As the newspaper, The Patriot, reported on June 26, 1968, following one of the raja’s frequent defections to an alternative coalition, leading to the formation of a new government, “The Raja who had been able to get his terms from Mr. Mahamaya Prasad [the former head of the Bihar government] assumed that he could demand from Mr. Paswan [the new head of the Bihar government] a higher price. This amounted to Deputy Chief Ministership and the Mines portfolio for himself and withdrawal of the innumerable cases filed against him and members of his family by the Bihar government.” ¹⁶ The raja understood that he could manipulate his bloc of backers to make and break governments and, in doing so, he could enrich himself a lot and help his followers a little bit in turn. That, indeed, is the lesson of bloc voting whether based on personal ties in Bihar, trade union membership among American teachers, tribal clans in Iraq, linguistic divisions in Belgium, or religion in Northern Ireland. Bloc leaders gain a lot, their members gain less, and the rest of society pays the price.

Bloc voting takes seemingly democratic institutions and makes them appear like publicly traded companies. Every voter or share has a nominal right to vote, but effectively all the power lies with a few key actors who can control the votes of large numbers of shares or deliver many votes from their village. Bloc voting makes nominally democratic systems with large coalitions function as if they are autocratic by making the number of influentials—that is, people whose choices actually matter—much smaller than the nominal selectorate of the rest of the voters. Since this is such an important aspect of winning elections we are obliged to explore how politicians do it.

The traditional approach has been to treat emerging democracies as patronage systems in which politicians deliver small bribes to individual voters. The New York Times, for instance, reported on September 17, 2010, in an article with the headline “Afghan Votes Come Cheap, and Often in Bulk,” that the typical price paid for an Afghan voter’s support was about $5 or $6. But the article also noted that widespread vote fraud probably made vote buying unnecessary in any event.

The explanation for fraudulent electoral outcomes based on vote buying in exchange for patronage is simple, but it is also incomplete. First, parties don’t bribe enough people, and second, once in the voting booth, voters can renege. Historically parties used to issue their own ballots. For instance, your party might print a ballot on pink paper. In such a way, party representatives could check that those who took bribes voted with pink ballots. Although we could fill a whole book with the tricks parties use to monitor vote choices, the reality is that today votes are likely to be anonymous, at least in real democracies.

Bribing voters works far better at the bloc level. Suppose there are just three villages, and suppose a party, call it party A, negotiates with senior community figures in the villages and makes the following offer: if party A wins it will build a new hospital (or road, or pick up the trash, send police patrols, plow the snow, and so on) in the most supportive of the three villages. Once a village elder declares for party A, voters in that village can do little better than support party A, even if they don’t like it. The reality is that there are so many voters that the chance that any individual’s vote matters is inconsequential. Yet, voters are much more influential about where the hospital gets built or whose streets get swept than they are about who wins the election. To see why, consider the case where two or three of the village elders declare in favor of party A and most voters in these villages go along with them.

Consider the incentives of an individual voter. Since at least two of three villages have declared for party A, an alternative party is unlikely to win so an individual’s vote has little influence on the electoral outcome. Voting for party B is a waste of time. Yet the voter could influence where the hospital is built by turning out to vote for A. If everyone else supports A, but she does not, then her village gives one less vote for A than another village and so loses out on the hospital. If she votes for A, then her village has a shot at getting the hospital. In the extreme case, where absolutely everyone votes for party A, our voter would give up a one third chance of getting the hospital in her village if she did not vote for party A. Voters have little incentive but to go along with their village elders.

By rewarding supportive groups over others, individual voters are motivated to follow the choice of their group leader, be that a village elder, a ward organizer, a church leader, or a union boss. The real decisions are made by the group leaders who deliver blocs of votes. They are the true influentials. It is therefore unsurprising that it is common for the rewards to flow through them, so that they can take their cut, rather than go directly to the people. Milton Rakove describes the process of handing out rewards to different ethnic groups under Mayor Richard Daley’s party machine in Chicago in the early 1970s: “The machine co-opts those emerging leaders in the black and Spanishspeaking communities who are willing to cooperate; reallocates perquisites and prerogatives to the blacks and the Spanish speaking, taking them from ethnic groups such as the Jews and Germans, who do not support the machine as loyally as their fathers did. . . .”¹⁷

Of course, leaders can use sticks as well as carrots. Lee Kuan Yew ruled Singapore from 1959 until 1990, making him, we believe, the longest serving prime minister anywhere. His party, the People’s Action Party (PAP), dominated elections and that dominance was reinforced by the allocation of public housing, upon which most people in Singapore rely. Neighborhoods that fail to deliver PAP votes come election time found the provision and maintenance of housing cut off.¹⁸ In Zimbabwe, Robert Mugabe went one step further. In an operation called Murambatsvina (Operation Drive Out the Rubbish), he used bulldozers to demolish the houses and markets in neighborhoods that failed to support him in the 2005 election.

Ownership of a public company works in the same way as bloc voting. We could hold our shares in our own name and vote at stockholder meetings. However, except for a very wealthy few of us, our votes are inconsequential and turning up is burdensome. Thus we hold stock via mutual funds and pensions (there are tax and management reasons to do so too, but then think about who has the incentive to lobby for these regulations). These institutional investors, like village elders, are influential enough that CEOs court their support. But it is much cheaper to buy the loyalty of the institutional investor by private goods, such as fees for board membership, than it is to reward all the little investors he represents with great stock performance.

So what can a politician do when elections are fair and the risk of electoral defeat is rising? When an incumbent is at risk of electoral defeat, he can always mitigate that risk by redrawing the boundaries of the constituency to exclude opposition voters. That is to say, the district can be gerrymandered, although this opportunity only comes once in a while so it may come too late to save an unpopular incumbent. The practice of gerrymandering has made it such that the odds of being voted out of a US congressional seat are not that different from the odds of defeat faced by members of the Supreme Soviet under the Soviet Union’s one-party communist regime. And, while gerrymandering virtually ensures reelection, it also makes the voters in a congressional district happy. After all, the gerrymander means that they get the candidate favored by a majority in the district. If gerrymandering isn’t an option, then other rule changes can be instituted, such as prohibiting rallies—in the name, of course, of public safety.

Have a look at the map of Maryland’s 3rd Congressional district in Figure 3.1. Need any more be said about why, in many districts, one party always wins?

Leader Survival

Building a small coalition is key to survival. The smaller the number of people to whom a leader is beholden the easier it is for her to persist in office. Autocrats and democrats alike try to cull supporters. It remains very difficult to measure the size of coalitions precisely. However, if we arrange political systems into broad groups of autocracy and democracy, then we can compare the survival of different political leaders.

FIGURE 3.1 Maryland’s 3rd Congressional District

Figure 3.2 looks at the risk for democrats and autocrats of being replaced given different lengths of time that they have already been in office. On average, for instance, democrats who make it through the first six months in office have about a 43 percent chance of being out by the end of their second year; autocrats only have about a 29 percent chance of being ousted in the same amount of time.¹⁹ Making it to ten years, democrats are three times more likely to be replaced than their autocratic, small-coalition counterparts.

These simple comparisons, however, miss an interesting and important detail. Although autocrats survive longer, they find surviving the initial period in office particularly difficult. During their first half year they are nearly twice as likely to be deposed as their democratic counterparts. However, if they survive those first turbulent months, then they have a much better chance of staying in power than democrats. Those early months are difficult because they have not yet worked out where the money is, making them unreliable sources of wealth for their coalition, and they have yet to work out whose support they really need and who they can dump from their transitional coalition. But once autocrats have reshaped and purged their supporters, survival becomes easier. Democrats, in contrast, are constantly engaged in a battle for the best policy ideas to keep their large constituencies happy. As a result, although democrats survive the early months in office more easily (they get a honeymoon), the perpetual quest for good policy takes a toll, such that only 4 percent of democrats survive in office for ten or more years. Nearly three times as many autocrats manage to accomplish this feat, 11 percent.

FIGURE 3.2 The Risk of Ouster by Type of Government

Staying in power right after having come to power is tough, but a successful leader will seize power, then reshuffle the coalition that brought him there to redouble his strength. A smart leader sacks some early backers, replacing them with more reliable and cheaper supporters. But no matter how much he packs the coalition with his friends and supporters, they will not remain loyal unless he rewards them. And as we will see in the next chapter, rewards don’t come cheaply.