Review: Harrison, "Secret Leviathan: Secrecy and State Capacity under Soviet Communism"
why socialist economics needs liberal legal institutions
When I read Michael Ellman's Socialist Planning (very good, read it), my main takeaway was my increased estimate of the "political burden" on the planned economy: so many of the issues that arose for it were direct results of an authoritarian system that concentrated power in a small number of hands and then insulated them from useful criticism. Far from being a natural fit, as imagined by both right-liberals and tankies, the party-state and the planned economy were a distinctly bad one: closed government makes everything under the direct purview of the state weaker, which is particularly a problem if the whole economy is brought within the state sphere.
Mark Harrison is a Hoover Institution guy, so, one of those right-liberals, but his thesis here is actually broadly consonant with this: the Soviet state's obsessive concern with security, secrecy, and censorship made it systematically less capable of accomplishing its ends, whatever they may have been. How much do his arguments and evidence support the thesis - are they weaker or stronger than would be expected? Ultimately I think the evidence here is quite strong, but also - possibly mostly as a result of Harrison's reasons for finding all this interesting differing from my own - limited in scope: Harrison is most directly interested in the burden placed by secrecy on the security apparatus itself, rather than on the civilian planned economy as a whole. I'll be focusing this review on parts that are most relevant to my own concerns rather than his.
The argument in brief: Harrison pictures a sort of curvilinear relationship between state secrecy and state capacity, with some secrecy improving the ability of a state to carry out essential security functions (defense, espionage and counterespionage, law and order) but with secrecy and censorship also coming with costs that start producing net problems past a certain point. These downsides include the direct bureaucratic cost of keeping track of secrets and keeping them under wraps, the informational downsides to keeping potentially valuable and useful information on a need-to-know basis, and the ease with which secrecy rules could be exploited by officials for petty ends. He then notes the extreme forms of secrecy and censorship that existed within the Soviet Union: "first department" political commissar offices that existed within a signficant number of workplaces, Party "conspirativeness" norms, and a legal principle of reflexive secrecy (that is, that the existence of any given secret is itself a classified secret.)
There are a lot of interesting anecdotes in here, but the funniest and most horrifying comes early on: a revised security code of the late Stalin era made breach of secrecy a capital offense, but (following principles of reflexive security) the regulations in question were classified at a higher level of security than many of the bureaucrats who would have had to abide by it, making it a crime for them to violate the rules and a crime for their superiors to make them aware of them. Fortunately, the wise decision of almost everyone to do basically nothing until "requests for clarification" had been answered meant that this Kafkaesque affair was essentially bloodless, but still: a terrifying moment, and a vivid illustration of how a reputation for secrecy and making heads roll could themselves make a government weaker at its own tasks (perhaps "too vivid" and parable-like to fully believe, are my only hesitations.)
Harrison looks at the Lithuanian KGB offices and finds that essentially a third of their work involved meta-management of secrets - actually, likely more than a third, given his estimate that these meta-documents would be more likely to be discarded as of less lasting value - and he speculatively extends to that to the rest of the bureaucratic infrastructure (maybe a little less than a third because the KGB was especially secret, maybe a little more because of the survivorship bias int the data.) I'd love to embrace this as confirmation of my thesis, but it seems way too strong: surely the state security services have to be way more concerned with meta-management of secrets than Lightbulb Factory #427.
Harrison also looks at the Soviet attempts to estimate their own defense budget - a sizeable proportion of the Soviet economy and hence relevant to civilian planning, but one that was almost impossible to estimate due to a combination of cascading telephone games (inevitable given the classified nature of almost everything military in nature) and implicit subsidies (that is, prices that were disproportionate to production costs and also disproportionately inputs into the military.) Here the latter speaks to different theses (and could be used to support a more traditional Austrian argument about calculability and price formation) but also is far from exclusive to planned economies (how much does the highway system in the US subsidize the military, this indeed being the formal justification for creating in the first place...?) The former is pretty important, and as noted important to the civilian sector itself, but worth asking how generalizable it is beyond that.
Two of the most intriguing pieces of evidence come from papers cited: Lichter et al. on the economic effects of East German informants and Egorev et al. on media freedom and bureaucratic quality. Lichter finds that more informant-dense areas of the GDR have worse social trust decades later, and that these effects might explain half of the current income gap between former East and West - if this were to hold up, it would explain much of the growth trajectories before the restoration of capitalism (considering we're then not talking about effects decades later but ongoing.) Egorov et al. find that controlling for oil endowments, freer media make state institutions more responsive and high-quality, which makes sense a priori and again indicates the poor fit between dictatorship and nationalization.
(As of writing this review I haven't read through these papers themselves and so can't comment on them, but they're in the queue somewhere.)
One area not explicitly brought up but where I think this might make Harrison's thesis highly generalizable: among the kinds of state firms most likely to have "first departments" and onerous secrecy requirements were those operating with frontier technologies (much of which was folded into the military.) Much of what we now know about disappointing performance by planned economies after the initial industrialization drive concerns not so much allocational or technical but dynamic efficiency - the speed at which new technologies were introduced and implemented. If the frontier research sectors were those where secrecy burdens were greatest, that would explain a slower-going Solow theta - but that's the variable that generally explains everything.
Overall, very glad I read the book, with my main hesitations being that its generalizability is limited by its choice of case studies, I don't know how intensely the main anecdotes and secondary literature is cherry-picked.