Planned Economy: Death by Quota

You have probably heard the joke: a Soviet nail factory is given a quota by weight, so it makes a few enormous nails; switch the quota to a count, and it turns out a flood of tiny ones. Two things behind the joke need straightening out. First, the hundred ways a metric can die, meaning how many directions a clumsy assessment metric can be exploited from. Second, the evidence grade of the joke itself: it is too useful, so useful that it gets cited repeatedly as established history, when it is in fact a fable with a complete blank where the primary sources should be. What is solid is the body of work economists built behind it. One line at the core: a real mechanism does not mean a real story.

Two things to connect first. The cobra effect in B04 already inoculated you: that story of a colonial bounty on snakes leading citizens to breed snakes likewise has no reliable historical source. Mechanism established, story doubtful, with the archived version being the rat bounty in Hanoi. The K01 case library made this a discipline: fables can be told, but a citation must state that it is a fable. The nail factory is another occasion for that discipline.

The person who turned the joke into serious scholarship was Alec Nove. His 1958 paper "The Problem of 'Success Indicators' in Soviet Industry" systematically analysed what he called the success indicator problem. What happens when a superior assesses a factory using gross output value, or physical quantities like tons, units or square metres: aggregate metrics do as they please with variety, specification and quality, and whatever single dimension you assess is where the factory moves all its resources, letting every other dimension go where it will. The paired counterexample Nove recorded is more convincing than the nail joke, because it is the same product distorted in two opposite directions: window glass planned by weight gets made thick and heavy to make up the tonnage; switch the plan to square metres and it gets made too thin to make up the area. Same machine, and when the metric changes direction the defects change direction with it. This is what the nail joke is trying to say without saying it rigorously, and it is scholarship you can check.

Opposite directions are only one way to die. There is a more everyday one, variety collapse. Nove says aggregate metrics do as they please with variety, specification and quality, and on the shop floor that means: if only gross output value is assessed, the factory picks the few easiest, highest-value-per-unit product types and mass produces them, while unfashionable specifications and troublesome models nobody wants to make go unmade. So the same plan is fulfilled beautifully while the market has a glut of popular goods and a permanent shortage of unpopular ones. Anything not written into the metric is effectively sentenced to death.

The person who assembled all these ways to die was Berliner in Factory and Manager in the USSR in 1957, recording the full technical repertoire of plan gaming. First, hiding capacity: managers deliberately under-report true production capacity, keeping a safety factor in reserve so that overfulfilling this year does not get next year's quota raised by the superior. That is the ratchet effect, where doing better means being squeezed harder next year, so the rational choice is always to be only slightly better than the quota. Second, the end-of-month rush, in Russian shturmovshchina: coast most of the time, then concentrate work at the end of the month to make up the reported numbers, with quality falling apart in the sprint. Third, the dedicated expediter, in Russian tolkach: because planned supply of materials is never reliable, a factory keeps people whose job is to go elsewhere and wheedle raw materials loose. Fourth, the network of connections, blat: using favours and private exchange to plug the holes in the planning system. Put these together and you see what the hundred ways to die really are: a clumsy single metric gets exploited by this system through every crack at once, in capacity reporting, production rhythm, material logistics and personal networks. And these techniques are not a Soviet specialty. In the 2000s, England's NHS also assessed hospitals with numerical targets plus executive jobs on the line. Researchers Bevan and Hood compared the arrangement, and recognized at a glance that it bore an obvious resemblance to the Soviet system, right down to a replay of the end-of-month surge. Incidentally, this set of observations predates any law named after a person: Ridgway wrote a review in 1956 saying that Soviet researchers had already concluded by 1940 that no single indicator can avoid this class of problem.

One open question: the nail factory tells us that a single physical metric distorts in opposite directions, so the Soviet answer was to add more metrics to plug the gaps, assessing variety, quality and cost together. But the more metrics were added, the more cracks there were to game, and the repertoire Berliner recorded did not stop working as metrics multiplied, it flourished. So is multi-metric assessment the cure or a new source of disease, under what conditions do metrics genuinely check each other, and under what conditions do they merely give the cheater more options? G04 tackles this line head-on, but that decades-long Soviet experiment never gave a clean answer.

The one-line takeaway: a clumsy metric has a hundred ways to die and Soviet factories demonstrated each of them, but that famous giant nail is a cartoon and not history; mechanism real, story doubtful, and keep the two apart when citing.