Talking Shop

Politics gets all the attention, but the public service itself should be more widely discussed.

A big welcome to all my new subscribers who’ve been streaming in from Twitter. I hope you find this publication in equal parts entertaining, informative, and mildly concerning.

---

Great libraries can be filled with weighty tomes expounding the virtues or drawbacks of various policy platforms. Entire data centres could be devoted to managing the social media bickering of partisan politics. Politics is engaging, no doubt, and there’s consequently no shortage of media coverage. Meanwhile, the actual business of governance, management, and delivery within the civil service is almost entirely ignored.

Over a third of economic activity in the OECD is via the public sector, with combined annual spending of approximately $32 trillion USD. At this scale, even a 1% improvement in productivity would deliver $320b of additional value for citizens each year¹. This could be achieved without requiring any significant changes to government policy, or requiring major political tradeoffs. Yet despite the obvious benefits to improvement, the state sector itself rarely measures productivity².

We live in an age of personal computing, modern machinery, and the world’s at our fingertips. These have helped drive a 50% improvement in the value of people’s time since the mid-90s, a key driver behind the trends in improving prosperity, health, and wellbeing that we’ve previously talked about, even as the number of hours the average person works continues to decline³. Yet despite these advancements, public sector productivity across the likes of the UK, New Zealand, and Australia has been stagnant since the mid-90s.

Public services are critical to society. Even anarcho-libertarians agree there are certain natural monopolies that are best managed by the state. We can debate where private/public responsibilities lie, but if we all agree a well-functioning state sector is important, why isn’t more attention paid to how it's run?

Sir Humphrey Appleby, the embodiment of the civil service. Portrayed by Sir Nigel Hawthorne in the BBC sitcom ‘Yes, Minister’. The show remains one of the best introductions to the civil service yet written.

There are a wide range of issues that need to be addressed⁴, but part of the challenge of improving public services is the difficulty of bringing talented outsiders aboard to help with delivering anything new. The process is fraught with misunderstandings, lacks a shared lexicon, and depending on the project can take too long to bring them up to speed, in situations where there is only a narrow window of opportunity to deliver results.

Operating in the world of business, there are many people one can learn from and emulate. Pick an obviously successful performer in your field, and there will be books, videos, workshops, and incubators discussing how to deliver. Meanwhile, many of the best public servants will simply quietly retire, taking all their institutional knowledge with them. Only a handful of people will understand what those ‘Tolkeinic heroes’ achieved for the people who paid their salary, and few would know to read their memoirs. After all, in most cases, good public service is invisible - clean streets, efficient tax processes, user experiences that resolve quickly and let people get back to their lives.

Failure is a natural consequence of the lack of context. How, for example, can a well-meaning advisor design an effective course for new managers when they lack access to frank discussion about what the problems are, let alone how to overcome them?

Consequently, delivering anything new within the public sector is reliant on trust networks, and a slow process of onboarding where bright-eyed newcomers slam head-first into brick walls until they leave the public sector in disgust - unless there's somebody around who can promise them it gets easier, and help guide them through the process of how change is actually delivered. Otherwise, they'll be stuck in the Kafkaesque labyrinth of irrelevant strategic documents and formal processes that only exist to confirm what has already effectively been decided in a pre-meeting.

"Surely that's not really how it works?"

“Well, yes, the formal process has become so unworkable after having a thousand pieces grafted onto it over the years that most decisions are resolved informally before being put through the formal process for the sake of appearance/regulatory compliance.”

“Shouldn’t that be changed?”

“The system is a natural consequence of the incentive structure. Attempting to change it without changing the incentives will just make it worse. Let’s focus on delivering meaningful changes for people, and build from there.”

Lacking context for failure, good people leave and tell their friends not to bother. Over time, the sense accumulates within the collective subconsciousness of public service not being a noble calling, but as a vaguely disreputable quagmire. Better to avoid the whole sordid business, and pursue something with higher prestige and pay, like a career in finance.

Improving the public sector is important because:

1. Doing so would provide enormous value directly, as we discussed above.

2. The state sets the regulatory environment upon which private sector incentives are either aligned (or misaligned) with the public interest.

3. Functional institutions are critical to a peaceful and prosperous society.

Even this last point is sadly often missed. Many of us in the English-speaking world have been lucky enough to live our entire lives in a society governed by the rule of law, where crime rates are at historic lows, where we can freely trade and enter into complex financial relationships with complete strangers with the confidence that the state will enforce contracts and punish duplicity. We take for granted that clean drinking water flows from the tap, unfamiliar roads will be well-paved, and that corruption will be kept to an appropriate minimum⁵.

We’ve been drawing down the legacy of the achievements of generations past. Great swathes of infrastructure built during the post-war boom are starting to reach end of life. Meanwhile, high-context systems implicitly designed for long-tenured staff are increasingly strained as the Boomer generation retires, while their Millenial and Zoomer replacements stay an average of less than three years. The situation is not yet critical, but in light of large fiscal deficits, high public debt, and (most critically) our aging populations, decay is inevitable without regeneration.

Consequently, if subsequent generations are to enjoy the same quality of life that we have, then we need talented people to enter public service and be able to see clearly and operate effectively. Attacting talent requires demonstrating that there are actually plenty of interesting, challenging, and important problems for them⁶. Retaining them means deepening the discourse such that they have the context, tools, network to solve them.

An unproductive public sector is not inevitable. Singapore is often held up as an exemplar, but most nations can boast of high-competence institutions that quietly and efficiently deliver their services. Why should we not aim for this everywhere?

---

In other news, I’m pleased to announce that the Taiwanese Government has invited me to return (following my essay ‘Why We Need Taiwan’) in early 2024 for a three-month research sabbatical looking into their use of data and AI in the public sector. I still have some spare time to allocate, so please reach out if there are topics you’d like me to look into, or people you think would be worth meeting, and I’ll do my best to cover them.

1. This assumes that each dollar spent by government delivers one dollar of net value to citizens. Obviously, we’d prefer if public spending did better than merely breakeven, though unfortunately even this bar is not always achieved, especially once cost overruns are taken into account. The ‘Iron Law of Megaprojects’ (they’ll be overtime, overbudget, and underdeliver on expected benefits) is a well-studied phenomenon and partly arises from our tendency to think in terms of normally distributed risks. In reality, delays tend to lead to further delays, problems arising from imperfect information and planning create further downstream problems, and so on. Thus, we would be better served thinking in terms of power-law distributions, as we’ve discussed previously.

2. As a further example, Australia’s 2023 Productivity Report only covers the public sector in passing - such as on page 6 where we read “… if productivity growth in an expanding non-market [public] sector remains in line with its measured historical average of zero, it would represent an increasingly large drag on overall economy-wide productivity growth.” [emphasis mine]

The private sector has an easier time measuring productivity, as profit/loss shows up directly on the balance sheets while public sector outcomes tend to be more intangible and harder to measure. However just because something is harder to measure doesn’t mean it’s less important, or that less effort should be put into improving it!

3. A true account of hours worked per capita is even more dramatic, as most statistics you’ll find are hours worked per worker, but this doesn’t account for the fact that people worldwide are also staying longer in education, taking longer breaks from work, and retiring earlier.

4. The NZ Productivity Commission state sector summary report provides a good overview of some of the high-level issues.

5. The ‘ideal’ amount of fraud/corruption in a society is zero, but given there are exponentially increasing costs associated with stamping out corruption (e.g. compliance costs, opportunities to use sweeping powers against one’s political enemies), the ‘optimal’ amount of corruption that a society should accept is non-zero. Same story with the management of risk. For non-catastrophic risks, if your failure rate is zero, then you’re probably not taking enough risks.

6. Admittedly they’re usually buried a layer or two deep. That is to say, the problems are typically not ‘how do we fix this problem’, but rather ‘given all these fiscal and operational constraints, how do we deliver the solution while letting people down at a rate they can absorb?’