#832 1930 · British Colonial Office (Warren Fisher Reform), studied by Guo Xu · Government / public administration
A 1930 reform stripped one official of promotion power over colonial governors -- and their tax collection changed overnight
the problem
Officials who owed their career to one connected superior optimized for pleasing that person, not for the organization's actual goals, and nobody had hard evidence the gap was costing anything measurable
background
From the Colonial Office's founding in 1854 until 1930, the Secretary of State for the Colonies held sole personal discretion over which colonial governors were promoted to richer, higher-salaried postings across the British Empire's roughly 70 colonies. This was ordinary institutional politics, not an obvious scandal -- rewarding loyal, well-connected men for good relationships in London looked like normal patronage, the same kind of appointment practice that had governed public office through most of history, and nobody at the time had a way to measure whether it was actually costing the Empire anything.
Economist Guo Xu digitized over 3,000 volumes of colonial personnel and public-finance records spanning 1854-1966 to test this directly. He found that governors personally connected to their superior in London were promoted to higher-salaried colonies at a measurably higher rate -- a roughly 12.7% salary premium over unconnected governors before 1930 -- and, once in office, these connected governors gave out more tax exemptions, raised less revenue, and invested less than unconnected governors did. Currying favor with local elites and avoiding friction built the kind of reputation that got a governor promoted; raising revenue for the Empire was not what patronage was actually selecting for.
what everyone would do
Treat patronage as an unfairness problem to be addressed on principle -- appeal to the Secretary of State's judgment, or simply accept that senior appointments have always run on personal relationships and that's the cost of doing business in a large hierarchical institution, since nobody had hard evidence the practice was measurably hurting the Empire's actual performance.
what they saw
Guo Xu's research (and the reformers who acted on similar concerns in 1930) revealed that patronage wasn't just unfair in the abstract, it was a precise description of what connected officials were actually optimizing for -- their superior's approval, not the colony's fiscal health -- because promotion depended entirely on the first and not at all on the second. Stripping one person of discretionary promotion power directly targeted the incentive, not just the appearance of favoritism.
the move
In 1930, following the Warren Fisher Committee's report on the Colonial Office's appointment system, the Secretary of State's private secretary was stripped of direct control over senior colonial appointments, and recruitment was handed to a new Colonial Service Appointments Board -- a formal, independent body overseen by the UK civil service commission, replacing what the Warren Fisher report itself described as a system where the Secretary of State 'has the sole power, through his private secretary, over the selection of candidates.' The reform, later called the 'Magna Carta of the Colonial Service,' severed the direct link between a governor's personal relationship with his London superior and his career prospects.
why it works
Because promotion had depended entirely on personal connection to one decision-maker, connected governors had every reason to avoid friction with local elites (granting exemptions, going easy on revenue collection) since currying favor built the reputation that got them promoted, while unconnected governors, unable to rely on that channel, had less incentive to trade away revenue for goodwill. Once the Colonial Service Appointments Board replaced the Secretary of State's personal discretion with an independent, criteria-based process, connected governors no longer had a promotion path built on pleasing one person, and their fiscal behavior converged with unconnected governors' almost immediately -- the same individuals changed how they governed once what determined their career changed.
the payoff
Guo Xu's difference-in-differences analysis found the connected-governor salary premium, statistically significant before 1930, became indistinguishable from zero after the reform (Table 4 of the published paper) -- the same governors' fiscal behavior changed once promotion stopped depending on who they knew, with the fiscal performance gap between connected and unconnected governors closing alongside the promotion gap.
where it breaks
This fix depends on there being an independent, sufficiently well-designed alternative evaluation process to replace single-superior discretion with -- a formal board that itself becomes captured by the same relationships, or one lacking the information to judge performance accurately, reproduces the same distortion under new branding. It also requires the organization's actual goals (revenue generation, in this case) to be measurable well enough that an independent process can select for them; where performance genuinely can't be assessed except through personal, relationship-based judgment -- highly ambiguous or slow-to-manifest outcomes -- removing discretionary evaluation may just replace one flawed signal with a worse one.
what came after
The study, published in the American Economic Review in 2018, is cited as one of the few precisely measured, historical natural experiments isolating what patronage actually distorts: not fairness in the abstract, but the specific thing connected officials optimize for (pleasing a connected superior) instead of an organization's stated goals (raising revenue, in this case) -- a more transferable finding than the general observation that patronage is unfair.
references
- [1]The Costs of Patronage: Evidence from the British EmpireAmerican Economic Review, vol. 108, no. 11 (Guo Xu), 2018guoxu.org
- [2]The Costs of Patronage: Evidence from the British Empire (journal record)American Economic Association, 2018aeaweb.org