Measuring governance

Creating a measurement system for government performance and why it hasn't worked so far

One of the hardest problems to get my head around has been that government is essential and useful and the underpinning of all civilisation but absolutely frustrating to deal with. But first, a few statements on my biases:

  • My childhood was spent in India where I was vaguely aware that government was a thing, mostly made its presence felt through the endless strikes people went on, and knew of what it could be because Kerala (and my family) had its share of rosy eyed communists

  • Another part was spent in Singapore where I promptly forgot that the government was anything but a phenomenal benevolent entity that let me pay my tiny tiny tax bill on a credit card and never bothered me beyond that

  • Then I moved to the UK and immediately understood where at least part of India's dysfunction came from, where government was simultaneously passive-aggressive and combative, with a lovely class system designed to be useless enough to be absolutely infuriating

  • Along the way I also experienced the governmental love in USA (man you guys complain a lot), Nigeria (I have no words), Germany (also no words…) and Denmark (great gingerbread houses..).

Across all of these varied experiences I couldn't understand what the hell any of these entities were interested in. Their motivations were opaque, you could ascribe malice about as easily as incompetence, and if they measured success in annoyance you wouldn't be too far off.

When I was much younger I used to think GDP was the key KPI. Turns out there's some consensus that it kind of sucks in actually measuring growth and prosperity, but the clarity it provides as a single metric is still useful. Regardless of Bhutan's effort to make happiness that metric (which doesn't seem to have worked if the comparative life satisfaction chart is any indication) nobody else has changed their true north.

GDP itself is of course a bit of a shoehorned metric anyway. Simon Kuznets who propagated the use of GDP had written that this is an amalgam of a bunch of measurements and therefore only useful as an approximate estimate. Which, if you take literally, means that we shouldn't be hooting and hollering about 2.49% growth in GDP vs an expected 2.45%. It wasn't meant to be a KPI for the government. For instance it doesn't capture quality and/or diversity in goods and services. It doesn't count externalities - positive or negative. And it doesn't account for services that don't have an easily measurable output. Granted some of these things should work itself out within the numbers in a long enough run, but there's a famous saying by someone about the long run being useless.

Is there something else available? One's perhaps the Human Development Index, helpfully created by the UN. It also seems to be weirdly well correlated with GDP so the utility is questionable. After all healthier and happier people are usually more productive, and vice versa.

Another option is to use specific indices like the Gini coefficient to measure particular problems like income and wealth inequality. There's also broader indices like health and wealth distributions (to understand the population), capital to labour ratios (to assess imbalance in productive sectors), total factor productivity (look at the name, wouldn't you want to measure it?).

What do you do when you get overwhelmed by choices? The answer is straightforward, you look at the list of 143 local government KPIs to make things easier.

Or perhaps we can go full McKinsey and use a balanced scorecard. If nothing else it'll give a lot of work to keep on top of, which is of course a clear indication that progress is being made. Work done on management is naturally work done in better managing.

But even so, considering there have been countless number of these proposals, which is it that interacting with the government seems a particularly thankless task, and a particularly hostile one, in almost all countries?


Now, one place you would've thought to turn to in trying to figure out a North Star for measurement is the world of business. Surely if you're in the job of managing money and people and assets you'd have to get pretty good at it, right? Right. This works out just about as well as you'd imagine.

The corporate world went through a multitude of measurement mechanisms. They still do better than the government though, though perhaps only because they work on a smaller canvas. Despite having a pretty unambiguous few measures in their sales and profits, there's also the mental fatigue of creating yet more balanced scorecards to help figure out how to run their company.

There's also the issue that the sheer number of stories that gets constructed around the metrics is staggering. If you don't believe me, just turn on CNBC any random hour and hear proud readings of that days horoscope in full confidence.

But there's no reason it should be this way. The basic measures that a business is measured by are a fairly comprehensive set that provides clarity on the business performance, as long as you don't go crazy with balanced scorecards.

And so there's another option, to try and provide a dashboard much like the ones used by today's start-ups. Or at least the venture capital investors. Here, almost explicitly, the aim isn't to push each metric individually but to try and so things that a) highlight problems fast enough, b) provide transparency on how the business is doing, and c) have something to benchmark performance against.

My favourite examples all come from the cloud companies world as that's where I spend a bunch of my time. In this world, being recurring revenue businesses at its core, there's a bunch of similarities at least towards some levels of government. You're not gonna manage all $20 trillion GDP this way, but you can at least manage each city, or the occasional state or province, this way.

After all the best feature of government is that it’s the ultimate recurring revenue entity. Your customers have to pay you, they don’t have a choice, which is a level of control even Larry Ellison can only dream of. So here goes a first cut on how to build a government that functions like a cloud software company:

  • Annual Recurring Revenues - Tax revenues - both for individuals and companies

  • Non recurring revenues - One offs like the sale of an asset or a levy

  • Financing - Bond offerings etc

  • Net Revenue Retention - for income tax at least a clear indicator of whether people are getting richer. Same for companies also I suppose.

  • Churn - for income tax again it's an indicator of if people are getting poorer (and dropping below the minimum). Same for companies of course.

  • Revenue growth rate - proxy (post NRR) for whether you were successful in actually attracting talent

  • Cohort growth - by income levels great way to see which segments of the populace is doing well

  • COGS - everything you spend to deliver the services - streets, police, the damn DMV - problematic because not all services are delivered by the same entity, but surely it can be reasonably disaggregated

  • S&M - efforts to attract talent to the city - the going preference seems to be to essentially say "SF sucks" and cool people pour in

  • R&D - infrastructure projects and general capex improvements

  • G&A - admin spend to make all this happen

And a few non-numeric indicators:

  • NPS - always a good idea to know if the people in your city like being in your city. Same for the state. Less so for the country as people aren't that mobile, though still probably a good idea.

  • Gini - at a city level you have a mobile workforce at the very top deciles and very bottom. At a country level you have much lower mobility except at the very very top, if that. This seems more an empirical observation than a theoretical fact, but it's an interesting one to keep an eye on.

  • High sunk costs - moving to the city is like installing an ERP system. You're not gonna rip it out easy!

As a result of not reading Twitter carefully enough, turns out David Sacks had done a fantastic tweetstorm that captures part of this thought already a couple days ago.

Must be something in the air ... Though David's thread is particularly applicable at the city level there's no reason to believe that this shouldn't happen.

And in the interests of transparency and balance, here's something obvious from some school in Boston. Perhaps it needn't be said that the government has different incentives than businesses, but better to say it anyway. Politicians aren't graded on their success in growing GDP or tax revenues.

This is both the pro and con of the system. As long as you don't break anything (or nothing breaks when you're in office), and at least one thing goes right, you can be considered to be a genius. No wonder that it's a maelstrom of bad incentives. And no wonder transparency is the first casualty of governance.

The government is trying to satisfy a large number of metrics, several of them non-numeric, so there's likely to be a big gap between the existence of these metrics and reality of policy implementation. But would it be too much to hope for that there can possibly be a future where we can get a reasonable shot at basic transparency? As an added plus point it could also engender some decent competition too which would be really nice.

And on the flip side there's every likelihood that this can and will be implemented, through a wonderful web interface like with ACA, made even more byzantine through the interaction of multiple subcommittees, and added on to the mountain of metrics that currently (supposedly) exists to create the perfect level of incomprehensibility. As usual with new and better standards, it's not the addition that brings clarity, but simultaneous removal of older ones.

Is this self-serving? Possibly. The thing is that this will be a boon for the data-focused and tech-savvy workers and we'll make out ahead for sure. Is it going to be good for the rest? I think so, but I'd love for this to happen somewhere so there's some empirical evidence one way or another.

It's normally futile to hope against the rise of complexity and increase in opacity, but this is one place I'd like to hold out hope. And on the plus side, maybe I won't need to fill out so many random pieces of paperwork again, fearing the Damocles Sword that swings above my neck for every mistake.