VoxEU Column Economic history Politics and economics

Costing secrecy

Democracy often seems bureaucratic with high ‘transaction costs’, while autocracies seem to get things done at lower cost. This column discusses historical research that refutes this. It finds empirical support from Soviet archives for a political security/usability tradeoff. Regimes that are secure from public scrutiny tend to be more costly to operate.

From public finance to climate change, democracy looks to be in trouble. In many Western countries, political decisions are gridlocked while economic, social, and environmental imbalances accumulate. Our leaders juggle public opinion, private lobbies, and expert advice while trying to live within past promises and present legal obligations. The costs of reaching decisions are often high and sometimes prohibitive, leading us into democracy’s ‘do nothing zone’, where bargaining fails and the outcome is procrastination (Wintrobe 2000: 247-279).

To judge from appearances, autocrats have the advantage. They make decisions behind closed doors. They manage or deflect the scrutiny of voters, journalists, and judges. Chinese and Russian leaders look at us with pity and contempt. Our own public opinion is vulnerable to what David Runciman (2013) has called ‘dictator envy’.

Appearances can mislead, however. The business of autocracy is done out of sight. While we observe outcomes, the process is hidden from view – until there is regime change and the backstage record is opened up to investigation. It then transpires that doing secret business is not so simple. There is a political security/usability tradeoff – a more secure system is more costly to operate. In recent papers (Harrison 2013a, 2013b) I demonstrate that the tradeoff exists and I evaluate it where possible.

Soviet secrecy

Until it collapsed in 1991, the Soviet Union produced secret paperwork on a colossal scale. Its archives hold millions of files that document the working arrangements of dictatorship (Gregory and Harrison 2005). Most files are stamped ‘secret’ or ‘top secret’. The starting point of Soviet secrecy was the denial of the right to know. Every functionary was trained in a fundamental principle, ‘conspirativeness’ (Khlevniuk et al. 1995: 74-77) – no one should be consulted over a decision, or even informed of it, without specific authority. The outcome was the most secretive system of government yet devised.

Soviet records show that secretive government has high costs, hidden at the time because of secrecy itself. These costs were of many kinds. Transaction costs arose through two channels – one procedural and the other behavioural. First, leak-proof government depended on costly procedures designed to assure secrecy. Second, harsh penalisation for secrecy violations induced fear and mistrust, causing officials to change their behaviour in costly ways.

Procedural costs: a government transactions tax

In order to ensure the security of government communications, the Soviet bureaucracy developed a system to account for secret paperwork. The system (first described by Harrison 2013a) tracked every secret document from creation through transmission, storage, and incineration or consignment to an archive. The system is documented in the Soviet archives by correspondence ledgers, storage inventories, deeds of destruction, and responses to discrepancies and violations. Because these records lack intrinsic interest, their significance has been overlooked.

The huge quantity of such records implies that officials had to devote considerable efforts to complying with secrecy rules. In effect, conspirative politics imposed a secrecy tax on the turnover of government business. The burden was multiplied because records that accounted for secrets were also secret, and therefore had to account for themselves.

It is challenging to estimate the procedural burden because the raw data comprise millions of poorly catalogued paper records. A unique digital catalogue of the archive of a small regional bureaucracy, the KGB of Soviet Lithuania, prepared for the Hoover Institution Library and Archive, allows us to measure the burden of secret paperwork by counting key phrases in file descriptions.

Among the catalogued records are 668,000 pages of documentation in 3,434 files relating to KGB management from 1940 to the late 1980s. Figure 1 shows the number of files by category and by the year the file was opened. Three categories are reported: counter-insurgency, secret file management, and other activities. The figure suggests three phases. In the first phase (1,992 files from 1940 to 1953) the KGB was highly focused on counter-insurgency. The second phase (1,003 files from 1954 to 1982) saw the KGB turn to normal peacetime operations. From the final phase (1983 to 1991) 438 files were archived, but during this time normal file management broke down: most open files were shredded as the KGB left Lithuania, while many others that might normally have been discarded were retained.

Figure 1. Lithuania KGB management files, 1940 to 1991

Source: Harrison (2013a: 1042).

In the years from 1954 to 1982, secret file management contributed around one-third (34%) of the total documentation preserved – a larger share than any other peacetime activity that can be distinguished.

The figure of one-third suggests two further questions. First, as an estimate of the procedural burden of secrecy, might we extend it to the Soviet bureaucracy overall? This would require several identifying assumptions, ranging from weak (that the activities of the KGB in Soviet Lithuania were representative of the KGB as a whole; that files retained in the Lithuania KGB archive did not overrepresent the activity of accounting for secrets) to strong (that the KGB was not markedly more secretive than any other Soviet organisation) to completely speculative (that paperwork intensity was uniform across activities – including the activity of accounting for secrets).

Second, when the government allocates resources to the security of its own processes, is one-third a lot? Comparative research on this subject does not exist, and no public records allow a contemporaneous comparison across the front line of the Cold War. In a more recent year, 2010, the Departments of Defense, State, and Justice accounted for 95% of US Federal outlays on the protection of American secrets, and their outlays ran at 3.1% of the direct costs of the general activities concerned.

This comparison is highly preliminary. Still, it supports the idea of a security/usability tradeoff by suggesting that the Lithuania KGB’s peacetime secrecy burden of 34% was greater than the American secrecy burden more recently by an order of magnitude.

Behavioural costs: Risk management

A case study (Harrison 2013a) examines how behaviour changes when secrecy is unexpectedly increased. In the late 1930s the Soviet state entered its most secretive phase. A particular secret was the identity and location of the numerous forced labour facilities administered by the Gulag (interior ministry chief administration of labour camps).

Figure 2. Soviet labour camps of the Gulag, 1930 to 1961

Source: The International Association ‘Memorial’ at http://www.memorial.ru/first.html

In 1947 a new law abruptly enacted long prison terms for disclosing secrets by accident or neglect. This was Stalin’s initiative. His targets were officials and researchers who, in his view, were too ready to share government information and scientific findings with the Soviet Union’s former wartime allies. The law was launched by a campaign to crack down on violators. New, more inclusive lists of secrets were promulgated. The new regulations were so secret that many of those held responsible under them lacked clearance to be informed of their new duties. A wave of fear swept through the Soviet administration (described by Gorlizki and Khlevniuk 2005).

Does fear have real effects? A thick correspondence file in the extensive Gulag archive held on microfilm at the Hoover Institution yields the answer.

Every Soviet labour camp was managed as a business that received supplies, delivered products, and made and received payments through the state bank as prescribed in secret government plans. As secrecy became more intense, everyday business was impeded. In order to continue to turn over goods and money as government plans required, camp commanders now risked severe punishment for disclosing a secret to counterparties – the fact that they existed. When the camp placed orders with a state-owned seller to deliver food, fuel, or clothing, the commander could not disclose the camp’s location without violating the new law. The camp held a state bank account with funds earmarked to pay for supplies, but to gain access to it the commander had to disclose the account holder’s identity, breaking the law again.

Exact compliance with the law would have caused government business to break down. In fact business did not stop because, although gripped by anxiety, everyone worked around the issues. These responses diverted effort from planned activities. Camp commanders lobbied to pass responsibility upwards and get higher officials to collude in workarounds that would keep business going. In Moscow, Gulag officials launched consultations and working groups to design new arrangements to comply with the new law. In the following months and years many alternative proposals were advanced, but no one would take responsibility to approve them. Rather than solve the problem they kicked it down the road, got used to it, and learned to live with it.

This case study gives empirical support to the role of fear in the security/usability tradeoff. Stalin’s intention was to make Soviet rule more secure. His initiative also raised the costs of Soviet rule; it frightened managers so that they diverted effort and attention away from planned production towards buck-passing, workarounds, and procrastination.

Concluding remarks

Every society has political transaction costs. The costs of open societies are in the open. Autocracies operate in secret. This biases the available evidence. It is hard to document the transaction costs of autocratic rule, and easy to jump to the conclusion that autocratic systems are less costly to operate.

Only historical work in the archives of autocratic systems can evaluate this bias. The evidence reported here suggests that the costs of government business under a secretive autocracy were much higher than was visible at the time.

If secrecy was costly, it was pursued in the expectation of some corresponding benefit – most obviously, regime security for the incumbent ruler. From this perspective, costing secrecy helps us to place a lower bound on the ruler’s valuation of freedom from scrutiny and lobbying.


Gorlizki, Yoram and Oleg Khlevnyuk (2004), Cold Peace: Stalin and the Soviet Ruling Circle, 1945-1953, New York: Oxford University Press.

Gregory, Paul R and Mark Harrison (2005), “Allocation Under Dictatorship: Research in Stalin's Archives”, Journal of Economic Literature, 43(3): 721-761.

Harrison, Mark (2013a), “Accounting for Secrets”, Journal of Economic History, 73(4): 1017-1049.

Harrison, Mark (2013b), “Secrecy, Fear, and Transaction Costs: The Business of Soviet Forced Labour in the Early Cold War”, Europe-Asia Studies, 65(6): 1112-1135.

Khlevnyuk, O V, A V Kvashonkin, L P Kosheleva, and L A Rogovaya, eds. (1995), Stalinskoe Politbyuro v 30-e gody. Sbornik dokumentov, Moscow: AIRO-XX.

Runciman, David (2013), “The Trouble with Democracy”, The Guardian, 8 November.

Wintrobe, Ronald (2000), The Political Economy of Dictatorship, Cambridge: Cambridge University Press.

630 Reads