The Inauspicious Zone

Roughly one Chinese county in sixteen sits, at any given moment, in a direction its city's mayor has been told is unlucky for him personally. Those counties record about 2 percent lower GDP than the rest of the same city.

The finding comes from a paper by Justin J. Hong and Yuheng Zhao, and the reason it is worth an hour of anyone's Saturday is not the superstition. It is the identification strategy, and what that strategy implies about every other decision nobody has thought to audit.

How you measure a belief

The unfavourable directions are not self-reported. They are derived from each leader's birth information, which the authors pulled from official biographies in prefectural yearbooks, and converted into personal unlucky orientations with the help of expert astrologers. A compass is then laid over a map.

That produces something rare in economics: an assignment that is essentially random with respect to everything that ordinarily confounds this kind of question. A county does not choose its compass bearing from city hall. It does not become more or less industrious when a new mayor with a different birth month arrives. The mayor rotates; the geometry does not; the direction changes anyway.

So the paper is able to hold the place constant and vary only the private belief of the person allocating the money.

The mechanism is mundane, which is the point

Nothing supernatural is being claimed. The effect runs through reduced policy support and lower public investment in the disfavoured zone, and then compounds through ordinary economic behaviour: firm entry falls about 6 percent, the productivity of the firms that remain falls about 4 percent, and population drifts slightly away.

That second-order chain is the expensive part. A mayor who quietly declines to champion a district is not doing anything a subordinate could file a complaint about. But businesses read the signal, stop opening there, and the ones already there get worse. The belief costs almost nothing to hold and a great deal to act on, and the acting is invisible because it looks exactly like ordinary prioritisation.

Scaled up, the authors' back-of-the-envelope figure is a loss of at least 0.1 percent of Chinese GDP annually, sustained across two decades.

The detail that makes it credible

An earlier version of the work reports that the effect is driven by prefectural mayors, with the impact of party secretaries close to zero.

That asymmetry does more for the argument than the headline number does. If the result were an artifact of geography, or of some unobserved feature of counties that happen to lie north-east of a city centre, it would not care which official's birth chart was used to draw the line. It would show up for both. Instead it attaches to the specific individual whose beliefs are being tested and whose office controls investment allocation, and vanishes for the one sitting alongside him.

A finding that could easily have been a coincidence brought its own control group.

It is not a freak result, it is a sharp instance of a known one

The temptation is to file this under statistical novelty. The wider literature does not allow that. Yang Yao and Muyang Zhang, working with a city-leader linked dataset covering Chinese cities from 1994 to 2010, found that leaders differ significantly in their ability to generate local growth, and that those differences survive transitory shocks and are not explained by leaders being moved non-randomly.

Their identification rests on the same institutional quirk: Chinese local officials rotate between cities, which lets a researcher compare the same leader across different places and the same place across different leaders, holding national institutions constant. The question "do individual leaders move growth" has an established answer, and it is yes, by economically meaningful margins.

What the superstition paper adds is a named, measurable channel. The general literature says leader identity matters without being able to say much about which part of the leader matters. Here one specific, documented, entirely non-factual belief is isolated and priced.

What this is actually about

Read as a story about China, or about astrology, the paper is a curiosity with a number attached. Read as a story about institutions, it is something more uncomfortable.

Every organisation that allocates capital does so through people carrying priors that never appear in the model. Not superstitions, mostly. A view about a region formed on one bad trip. A discomfort with a category of business nobody has examined since it was acquired. An unexamined preference for the kind of proposal that resembles the last successful one. These are not corrupt and they are usually not even conscious. They are simply inputs that carry real weight and leave no record.

The Chinese case is legible only because the belief happened to be systematic, ancient, externally documented and mappable to a compass. Almost none of the equivalent biases in a Western firm have any of those properties. They are idiosyncratic, undocumented, and would require knowing something private about a specific executive to reconstruct. Which means the measured 2 percent is not an upper bound on this class of problem. It is the one instance where the problem happened to be measurable.

Why nobody finds their own

The structural obstacle is not analytical. It is that the people best positioned to detect a leader's systematic bias are the people who report to that leader.

A subordinate who assembles evidence that the boss consistently underweights a region, a product line or a class of customer has produced a career risk, not an insight. The analysis is straightforward, and in most organisations the data already exists in approvals, budgets and headcount by unit. What is missing is anyone with both the access and the standing to run it and survive the answer.

This is precisely why the finding required outside academics. Hong and Zhao had no stake in any Chinese mayor's reputation, no promotion to lose, and access to a public record of birth dates. Every one of those conditions was necessary.

The question worth sitting with

The practical version is not "what are my biases," which invites an honest but useless answer. It is narrower and harder: if a systematic, expensive allocation bias existed in this organisation, through what mechanism would it be discovered, and by whom?

Most organisations, asked plainly, do not have an answer. Their audit functions test whether decisions followed process, not whether the pattern of decisions has a shape nobody intended. Their board reporting aggregates to a level at which a 2 percent regional shortfall is invisible. Their analysts work for the person whose pattern would be under examination.

What the Chinese mayors had, uniquely and accidentally, was a foreign researcher with a compass and no reason to look away. It is worth asking what the equivalent would be in any given firm, and worth noticing how quickly the honest answer becomes "nobody, and there is no process by which anybody would."

Six percent of counties. Two percent of output. Twenty years. Found by strangers, because only strangers were free to look.

Sources

• Justin J. Hong and Yuheng Zhao, "The Costs of Leader Biases: Evidence from Superstitious Chinese Mayors". Source for the approximately 2 percent lower GDP in zones perceived as unfavourable to a mayor, the mechanism of reduced policy support and public investment, the 6 percent fall in firm entry and 4 percent fall in the productivity of remaining firms, the small population decline, the share of counties falling in an unfavourable zone, and the back-of-the-envelope estimate of at least 0.1 percent of annual Chinese GDP over two decades.
• "Leader Biases and Growth: Evidence from Spatial Superstitions", Chinese Economists Society. An earlier presentation of the same research, and the source for the derivation of unfavourable orientations from leaders' birth information with the assistance of expert astrologers, for the birth details being taken from official prefectural yearbook biographies, and for the finding that the effect is driven by prefectural mayors while the impact of party secretaries is close to zero. The two versions differ slightly in the reported shares of affected counties and in the headline effect size, which the earlier one gives as a 2 to 3 percent range; we have used the later paper's figures throughout and flag the discrepancy rather than reconcile it silently.
• Yang Yao and Muyang Zhang, "Subnational leaders and economic growth: evidence from Chinese cities", Journal of Economic Growth, volume 20, 2015. Source for the city-leader linked dataset covering 1994 to 2010, for the finding that leaders differ significantly in their ability to promote local growth, for the rotation-based identification that isolates leaders from national institutions, and for those estimates being robust to transitory shocks and to non-random moves.
• Justin Jihao Hong's research page and the HKU Business School seminar listing, given as routes to the paper: the SSRN record above refuses automated requests and may not open from every reader's connection.
• The Economist, issue of 8 to 14 August 2026, for surfacing the research. It renders the effect as 2.3 percent; the paper itself reports approximately 2 percent, and we have used the paper.