What, exactly, causes the tradeoff to bind? There are three factors. First, as secrecy becomes more and more intense, the government must commit increasing resources to managing the “regime of secrecy” and maintaining its integrity. These resources must be taken from elsewhere, and the diversion places an increasing burden on state capacity.
A second factor that comes into play as secrecy grows is not less important, but it is more complicated: the bundling of functions. Secrecy enables some functions and disables others in bundles that cannot be disentangled. An example is the bundling of error correction with the management of authority (or reputation). Suppose a policy has a bad outcome: the excessive rapidity of Soviet industrialization in the late 1920s and early 1930s led directly to a famine in which millions of people starved to death. To maintain the authority of the party and its leaders, this bad outcome was strictly censored. Because of this, private citizens and most public officials were deprived of information that could have been useful in preventing a repetition. Thus, in enabling the management of authority, secrecy also disabled error correction.
Evidence in support of this bundling is provided by economists Egorov, Guriev, and Sonin: using data from the period 1995-2007 and controlling for income levels, democracy scores, and natural resources, they show that bureaucratic quality across more than 130 countries was positively and significantly increased by the degree of media freedom.[114] The authoritarian ruler’s problem is that media freedom opens a channel to the improvement of policies and their delivery, but the same channel can allow criticism to spread and opponents to coordinate action.
Another example is the bundling of concealment (from an external adversary) with cooperation (internally among managers responsible for the business of the state). Suppose the location of a sensitive facility is encrypted, with the key kept under secrecy. Outsiders who do not have the key can no longer find the facility on the map. But insiders—say, managers with legitimate business at the secret facility—who lack the key, cannot find it either. In this real-life example (from evidence to be provided in Chapter 4), secrecy switched concealment on and cooperation off at the same time.
And a third factor in the secrecy/capacity tradeoff: As secrecy becomes more intense, it provides a cover for those who choose to exploit it, not in the interests of the state, but because they are greedy or lazy. You want to inspect my work? No, it’s secret. You want to know how much I’m entitled to charge you? No, that’s secret too.
The sense of this discussion is captured in Figure 2.3. This figure shows state capacity (increasing vertically) as a function of the intensity of figure 2.3. The secrecy/capacity tradeoff: State capacity rises, peaks, and then falls away as the degree of secrecy increases
Note: In this figure the degree of secrecy (defined as increasing regulation of information and penalization of violations) is measured on the horizontal scale. State capacity is measured on the vertical scale. Cmax marks the maximum attainable level of state capacity, and S* is the degree of secrecy that delivers the maximum at point X. This chapter argues that citizens of a democracy are likely to prefer combinations to the left of the maximum, such as Y, whereas an authoritarian ruler is likely to prefer a combination to the right, such as Z.
secrecy (increasing horizontally from left to right). Secrecy is made more intense by increased regulation and increased penalization of noncompliance. By assumption, secrecy has diminishing returns and increasing costs. At low intensities, the costs of secrecy are low and the benefits are high. The rising part of the curve reflects those aspects of secrecy that promote state capacity by providing minimal assurance of the power to take decisions without being manipulated by foreign and domestic adversaries. While marginal costs are low and benefits are high, state capacity rises. As costs increase and benefits diminish, a peak is reached at X, after which a decline sets in. As secrecy becomes even more intense, state capacity is driven down by the costs of assuring the growing volume and detail of secrets, the difficulty of protecting the business of the state from the growing complications of secrecy, and the growing scope for insiders to exploit secrecy for personal gain.[115]
As this approach hints, somewhere in the figure there should be a “right” level of secrecy—but where? That very question was assigned to the journalist and former spy Mark Frankland during his training for the British Secret Intelligence Service in 1958:
We had to write an essay on “the economic level of security.” Total security led to paralysis, for only if no one said or did anything could nothing be given away. Too little security led to disaster. Where was the golden mean? It was as good, or as pointless, a subject for debate as the medieval argument about how many angels could sit on the head of a pin.[116]
Although seemingly scholastic, the location of the golden mean, the “right” level of secrecy, is a practical issue for all states, one that is vital to get at least approximately right. But where is it to be found? It would be easy, but wrong, to identify the golden mean with the point at X, where state capacity finds its peak.
A moment’s reflection suggests the mistake. The answer depends on the constitution of the state. In a liberal democracy, where an “informed citizenry” is valued, the voters might tend to see secrecy as a necessary evil. If so, they should be willing to give away some state capacity so as to have less secrecy, preferring Y to X, for example. A degree of openness to foreign influences, whether welcome or unwelcome, or an element of domestic gridlock resulting from political competition, might be a price worth paying for the right of citizens to be fully involved in the making of decisions or to challenge the authority of the decision makers.
For the dictator, by contrast, secrecy is not a necessary evil but a good. Such a ruler should be willing to give away some state capacity so as to have more secrecy. Beyond X, where secrecy is beginning to erode state capacity, lies a point such as Z. Here the ruler willingly sacrifices some capabilities in order to conceal mistakes and wrongdoing, maintain a façade of competence, and suppress criticism and opposition.
In the case of communist parties in power, it appears that they wanted two things: to secure their own regime, and to build a powerful state. At first sight, these two objectives looked fully compatible. Looking more closely, the notion of the secrecy/capacity tradeoff drives a wedge between them. It suggests that secrecy promoted both regime security and state building up to a point. Beyond that point, increasing secrecy continued to fortify the defenses of the ruling party, but it began to work against the goals of state building. This was because the costs of secrecy began to eat away the capacity of the state to do anything more than keep its own secrets and keep the ruling party in power.
The idea that a dictator faces tradeoffs is not new. Any authoritarian ruler must spend resources on repressing opponents or on buying them off. These resources must be diverted from other uses, which must be traded off.[117] Again, the more fearsome the dictator, the less he can trust the sycophants that surround him to share unwelcome facts: this suggests he must trade truth for compliance.[118] Another recognized tradeoff is competence of government personnel for loyalty: autocratic regimes are likely to promote their officials based on unthinking loyalty and adherence to the leader, to the detriment of their general level of competence.[119] The contribution of this chapter is to identify secrecy as one of the margins at which the dictator must optimize.
114
Egorov, Guriev, and Sonin, “Why Resource-Poor Dictators Allow Freer Media,” 660-61. Their study used scores for media freedom from Freedom House and government effectiveness and regulatory quality scores from the World Bank; it used instrumental variables to check for reverse causation.
115
Economists will recognize this as a “toy” model in the sense of Bliss, “Application of Toy Economic Models.” It lacks generality; it is “silly, little, and useful.”
118
Wintrobe, Political Economy of Dictatorship, 20-39; Greitens, Dictators and their Secret Police, 34.
119
Egorov and Sonin, “Dictators and Their Viziers”; Zakharov, “Loyalty- Competence Trade-Off in Dictatorships.” The prehistory of this idea is discussed in Chapter 5.