In the coming months there will be many conversations about what governments everywhere can learn from COVID-19, and what they can do in future to be better prepared for crises.
The list will — I hope — include very wide questions about social resilience, health infrastructure, internet infrastructure, monetary policy, housing, land, and the structure of our economy. It will also some very specific, practical questions about government resilience: building a fully digital, secure government, so any team in any part of government can pick up their devices and relocate in a few hours without interruption, and without compromising the security, privacy or democratic accountability of government services or functions.
But before we can do any of those things, there is one system that we need to fix first. One that is so boring, so invisible, that it could easily get missed: procurement. That is, how government buys things.
What have we learned from COVID-19?
As the outbreak became a pandemic, governments realised they would need to increase the capacity of their health systems to deal with a specific need very, very fast. That meant the NHS was going to need tens of thousands more ventilators, several more hospitals, millions more items of PPE, testing kits, meals and so on, and quickly. The catch was, the normal way of getting those things isn’t scaleable. Ordinarily they are probably made steadily by one or two companies here in the UK, or, more likely, they are manufactured in China. The obvious problem in the latter case being that every country on earth had more or less exactly the same need at the same time. In a crisis, simply relying on being the richest person in the room is either useless, immoral or both. There are still, it turns out, some things that money can’t buy.
However, at the same time, across the UK there were thousands of small and medium enterprises (SMEs), from fashion businesses, manufacturers, constructors, food businesses, testing labs, 3D printing businesses and so on who suddenly found themselves unable to carry on business as usual, but are willing and able to switch over their product lines and contribute to the effort instead. I’ve seen and heard of so many examples in recent weeks and days, such as Makerversity, who rapidly set up a micro manufacturing network called ‘Shield’ to produce masks, face-shields, gloves and a variety of PPE.
Clearly the will – and the capability – to pull-off a ‘Dunkirk’ style effort is there; where the collective power of ‘many small’ can rise to meet a big challenge. It never ceases to astound me what can be achieved when many smart, passionate people are aligned around a common goal. It is something we rarely see outside wars, space races or crises. COVID-19 is one such crisis.
And yet, the Shield initiative — and many other SMEs — have found it extremely hard to get their products to where they are needed, let alone paid for. In many cases they’ve ended up finding other buyers, or giving them away and relying on donations. Where rapid deals have successfully been made by government, it seems to have mostly been with a small number of large providers, such as Dyson, McLaren F1 and Burberry, but below this scale, it seems to be much more difficult.
So what’s going wrong? While there will be lots of criticism suggesting that this is a problem of political competence or will — and perhaps some of it will be valid — I suspect there is actually a more structural problem at play here: government literally does not know how to procure from a thousand SMEs. And it certainly doesn’t know how to do it rapidly.
Procurement needed fixing anyway
Of course, we didn’t need COVID-19 or ferry companies with no ships to tell us that our procurement systems are broken. We’ve known it for some time. From buildings, to infrastructure, to machine parts, to waste collection services, to IT, at some point all of us have wondered why it is that whenever government tries to buy anything it ends up being weirdly expensive, slow and quite often, a bit rubbish.
What I’m going to suggest is that (for the most part) this is not because individuals in the system are unusually incompetent or malign, but rather it is a function of the way that our 20th century procurement systems were designed.
Government procurement is a hugely complex topic, and many government reviews and white papers have been written by people far more expert than I. But at its core, it is founded on some pretty basic fundamental tenets.
First, the idea that if you want the best price, you need market competition, and if you want market competition, you need some system of bidding. This has always been true, and will remain true.
Second, the idea that bigger is better. The bigger the deal, the better the price, because a bigger buyer has more negotiating power and bigger suppliers can invest more into an efficient supply chain, so can produce products and services more efficiently. This was once true, but in the 21st century, technologies like the web and low-cost digital fabrication tools may prove it to be only sometimes true, and often false. More on that later.
Third, because it involves public officials making decisions on behalf of the public purse, there is a high risk of corruption (either explicit, where favours or envelopes of money change hands, or tacit, because the public official goes with whichever option is more convenient for them personally, or that benefits a friend of a friend etc.) In economics this is called the ‘Principal Agent’ problem. It makes it necessary to put very tight, standardised, anti-corruption measures in place.
All this seems pretty reasonable. So what’s the problem?
Well, I’d suggest there are three. Let’s call them the ‘closed market’ problem, the ‘innovation’ problem and the ‘bandwidth’ problem.
The ‘Closed Market’ problem
In recent decades, there has been a tendency to refer to a thing called ‘The Market’ without really questioning what that means. Usually, in public procurement when people say ‘The Market’ what they actually mean is ‘The small handful of large companies who are the existing suppliers’. By Adam Smith’s definition, this is actually closer to the opposite of a market.
The root of the ‘closed market’ problem lies in the question of what it is exactly that the government is buying. Are they buying just the product or service? Or are they also, in-effect, procuring its R&D? Are they procuring a package that involves setting up a whole new supply chain from scratch? Are they procuring finance for the product (eg, ‘Private Finance Initiative’ (PFI) contracts)? Are they outsourcing risk (eg. ‘Design & Build’ style construction contracts)? In recent decades, the answer to this question has very often been: all of the above. Outsource everything, it’s just easier that way.
The problem with bundling everything together like this is that these contracts become such huge undertakings that only big players, with an established ‘track record’ can successfully bid for them. It’s pretty much impossible for SMEs and innovative new market entrants to break-in. This means the so-called ‘market’ becomes a small cartel of huge companies hoovering up all the contracts. There is plenty of scope for tacit collusion between this closed circle of companies, to keep prices high.
A second side-effect of these high-capital-cost-contracts is that bidders demand long contract durations to recoup their initial investment; sometimes even 10 years or more. So sure, there is a brief phase of market competition during the tendering phase, right up until the moment the signatures are put onto the contract. But at that point, the winning company (often the lowest bidder) has effectively won a 10-year monopoly. From the moment the ink hits the page, the supplier has a direct, rational incentive to minimise the quality of the product or service (think of those crumbling PFI schools), and to spend zero on innovation that would benefit the customer, society, the economy or the environment. Their professional duty is to maximise returns to their shareholders, and that means keeping the product or service as bad as they can get away with. For example, everyone knows that we need to radically transform our waste system in the UK today, but the companies contracted to deliver waste collection and disposal services have little or no incentive to do it. This is not because their staff or executives are somehow greedy or don’t care – they’re good people, like anyone else – but the business model we have created means they have a duty not to do it.
Worse still, the suppliers have an irresistibly strong incentive to design the product or service in such a way as to make it as difficult and as expensive as possible for their customer to switch to a competitor when the contract does come to an end. Take, for example, software contracts, where suppliers often try to hold intellectual property rights over essential functions or public data registers; or make their software as un-interoperable as possible in order to make customers pay extra for ‘connectors’; or make it as difficult as possible for customers to recover their own data and port it over to a new system. The combined effect is that, at any given moment in time, most public bodies are working with software that is at least 10 years out of date. That, in turn, creates even further dependence on outsourced services, and results in failures like the Wannacry ransomware attack, which cost the NHS £92m.
The great irony of this ‘big’ public procurement model is that government becomes so utterly dependent on these huge companies that they become ‘too big to fail’. So when they do, government ends up having to bail them out or pick up the cost anyway.
An obvious example was the contractor Carillion, which in truth was less a construction firm than a financial vehicle for outsourcing risk (the actual construction work was done by subcontractors). Shareholders and executives extracted significant profits and bonuses from the company, before leaving the taxpayer to pick up a £148m bill. So, in the end, it turns out it was the taxpayer who was carrying the risk all along.
(Update 12th May 2020: There is also an interlinked problem here that we could call ‘the public goods capture’ problem, where large companies use artificially low bids to capture a lock-in, or to gain access to public goods that have far greater value than the contract itself (such as data about citizens). This was exemplified recently when Palantir won an NHS bid to handle the COVID19 app data by offering to do the work for only £1. We don’t know precisely what it is they are aiming to capture, but the transaction itself should ring alarm bells: it is hardly likely to be a gesture of altruism on Palantir’s part, but rather a smart strategic move to play the system.)
The ‘Innovation’ Problem
The ‘closed market’ problem is already a bad enough innovation problem, but there is also another one.
Many ideas for how to improve government or public infrastructure don’t come from within the public sector. They come from citizens, businesses, or social entrepreneurs.
Connecting these people with those in the public sector with proximity-to, knowledge-of and responsibility-for public problems is already hard enough. But even if they do manage to find each other, and even if they can find budget to do it, they then discover that they have another problem. How to buy it?
Procurement rules mean that above a certain threshold, an open tender must be put out, looking for at least three bids. At this point, one of the existing incumbent companies with a long track record can offer to deliver a terrible version of the proposal, stripped of all social, economic or environmental value (all the stuff that is difficult to account for), on terrible terms, but at a low price. Unless the original proposers can measure the immeasurable, and prove the unprovable, then the public authority must choose the cheaper of the bids. The original proposal is hijacked and taken on a race to the bottom.
And so it is that procurement rules that were originally intended to prevent corruption can end up creating Kafka-esque levels of bureaucratic absurdity and common-sense failure. The explanation can’t be laid at the door of any one person’s actions, or any one point of failure; rather it is the compound effect of many independently reasonable actions, whose effects cascade to become collectively catastrophic.
In this situation, the only option available to the public official is to try to cheat the system, by writing a tender that is so specific to the original proposer than almost no one else can beat them. But if it works, all it demonstrates is that the rules intended to prevent corruption are ineffective: they can be gamed. The same trick could just as easily be used to fund projects that do not represent public value. (Some would argue that this is exactly what happened in the case of London’s ill-fated ‘Garden Bridge’).
This stifling effect on innovation also has the side effect of lending ammunition to those who would argue that we should ‘cut red tape’ by cutting regulations or relaxing anti-corruption protection measures. Very bad idea.
The ‘Bandwidth’ problem
The third problem – bandwidth — is very evident in the case of providing PPE to the NHS and care workers, but it shows up in many other forms of procurement too, such as affordable housing.
Put simply, buyers only have a finite mental bandwidth, finite networks and finite capacity to work through the administrative overheads of tendering. Especially today, since many of the processes involved are still very analogue and communication-intensive. Every tender has to be designed, checked, put out for comment, and then a time window created for questions and responses from suppliers. These all add up to turn a decision that a private sector executive could make in a few hours into the business of weeks.
These high transaction costs mean that it remains much, much easier to buy 1000 units from 1 person than it is to buy 1 unit each from 1000 people.
As a result, huge potential capacity in the ‘long tail’ is left un-tapped, and public money that could be going to local SMEs flows to large contractors and their shareholders.
How do you solve a problem like procurement?
So, if we were to redesign procurement for the 21st century, to try to imagine a system that is faster, cheaper, more effective, more competitive, more open, gets better public value and provides better protection against corruption, what might that look like?
Well, I don’t know. But if you’ve read this far, you’re probably interested enough in this problem that you might have some ideas. These are mine:
- Digital platforms
The first and most obvious move is to use the web to solve the bandwidth problem. It’s what the web does best. Government can learn from the effectiveness of platforms like Amazon, Ebay, AirBnB, Alibaba and the App store; all made possible by standardised template agreements, simple protocols and well-designed digital interfaces that aim to get the transaction costs as close as possible to zero. The Digital Marketplace and GCloud framework are a first step towards this. The objective should be to create truly open markets – that is, markets where any new entrant can join and quickly prove themselves capable of delivering. As far as possible, these could shift from filling in long application documents, towards a more ‘ratings’ like approach, where a supplier can easily get a small first order and prove themselves. The real proof is always in the pudding, and even the smallest portion tells you more than whole books of recipes. Right now, a busy, stressed, NHS care trust executive should be able to get online, and order 200 face shields each from 10 companies, in less than 10 minutes.
However, what makes web platforms effective is not just the frictionless digital infrastructure, that lets buyers connect with sellers and vice versa. Their success lies in the power of the platform owner to shape the market: to set standards and enforce rules of engagement. Uber do it. What if government did too?
2. Rules-based procurement
This ability to aggregate the public sector’s combined purchasing power to set the rules of the market is the single most powerful tool that government is not using.
Let’s return to the example I mentioned earlier, of IT contracts. A year or so ago, in collaboration with Connected Places Catapult, Tech UK and with contributions from a range of experts and suppliers, we drafted a checklist of 15 simple but robust basic checks for any local government IT contract. It included things like a customer’s right to extract their own data at any time, and a ban on any contract longer than 2 years. We realised that if government were to make all public sector IT contracts conditional on meeting those 15 rules, government could prevent most, if not all, of the kinds of market failure we see across public sector IT every day, where government ends up paying millions for rubbish, outdated software. In other words, there is no such thing as ‘The Market’. There are many possible markets, and government has more power than it thinks it has to shape markets where it is almost impossible for them to go too far wrong. I would argue that not only does government have an opportunity to use its platform power more effectively, it in fact has a moral obligation to do so.
Of course, there were some (not as many as you might think, I should add) murmurs from the old suppliers who said ‘if you make the rules of the game too stringent, we won’t play’. But that’s the whole point of competition. If the old companies aren’t willing to play, they leave the market open to younger , smaller, hungrier, more innovative companies who will. And believe me, they will. I’ve written about this before under the concept of ‘Democracy as a Platform’.
(Additional note: In other domains, this could go further. For example, in cases where a contract will inevitably create a temporary monopoly, or where there is a particular urgency to move fast in a crisis, bids could be limited to non-profit companies (no profits to shareholders) with wage ceilings.)
3. Open, modular contracting
Another approach we’ve been exploring, particularly in relation to procurement of buildings, is to shift away from the practice of bundling everything together (R&D, finance, delivery, risk etc) into a massive contractural black box, and instead doing the exact opposite: breaking everything you want to buy into small, separate, predictable, transparent modules, each of which is documented for all to see, and can be procured from any (or several) of a range of suppliers at any time.
Let me give an example from our work on WikiHouse. In recent years, if a council wanted to build social housing on a site, they might go to a single contractor to do the whole thing. Of course, there are a limited number of construction firms capable of taking on a task and risk of this scale, so the bids come in high. All the design IP, costings, task data etc is black-boxed, kept by each contractor whose costs are largely based on guesswork. So when the next project comes along, nothing gets much better, cheaper or more predictable. Councils are stuck in groundhog day, re-tendering to the same group of construction firms, and starting every project from scratch.
Digital manufacturing allows us to change this, because it allows construction to be broken into clear, predictable tasks, that almost anyone can do.
So what if, instead of going out to tender, the council were to start by paying an R&D team to develop a customisable house type design, with an IP agreement that allows them to use that design as many times as they please (this might be a open source, or a commercial licence). Every component, every task is predictable and separately documented, all recorded in spreadsheets and assembly manuals that everyone can see. Prospective suppliers (who might be small shop-fitting companies down the road) then provide rules of thumb for the costing of each task (eg ‘£200 per day’ or ‘£21 per sheet of plywood’). These rules of thumb are their ‘bid’ that can be used to cost a task. If the customer offers them the job, they don’t get to change their bid, they only get to say ‘yes’ or ‘no’. If they do accept, they are obliged to measure as they go and feed–back into the design and documentation, so the predictability of cost and timescales gets better and better over time, for everyone to see. This also means that instead of being stuck with the same old large suppliers, the council can grow an open network of SMEs capable of delivering the same product or service, and switch between them at any time. Continuous competition, continuous collaboration and continuous agile innovation.
This approach has the additional benefit that, by taking on additional risk (and then using feedback to reduce it) the customer also regains control over design quality. The contractors’ incentive to drive-down quality has been politely removed from the equation.
This approach may not work for everything, but I suspect it could be applied to a huge variety of procurement tasks. It could certainly be used in the case of the COVID-19 PPE.
5. Flat pricing
The final idea — which certainly won’t apply to every situation, but might apply to some — involves flipping the game of bidding itself upside-down.
The traditional contracting mindset is that you invite bids on price from different suppliers in order to ‘let the market find the price point’. You then pick one (often the second-to-lowest bidder). The trouble with this, as we have already explored, is that it creates a strong incentive for suppliers to bid low to win the job, then immediately engage in a race-to-the-bottom on quality or resilience. In the end, the money saved by getting suppliers to bid-low often turns out to be a false saving.
But in many situations it may increasingly become possible to use shared data to predict roughly what the price point should be, or what the customer can viably afford for it to be. So what if, instead of getting suppliers to bid on price, we were to fix the price across all potential suppliers, but then invite them to outbid each other on quality or performance. In effect, to trigger a race to the top on quality and other social or economic outcomes. It’s a way of extending an open invitation to all possible suppliers, including new entrants, and saying ‘here’s the money that’s on offer, what could you do with it’?
Of course, if no suppliers step forward, then the price would have to be raised until several companies offer their services.
Let’s fix procurement as if our lives depend on it. Because they do.
Or perhaps, all these ideas are useless, and it’s another idea entirely. But let’s talk about it. In fact let’s do more than talk about it. Let’s start prototyping, piloting and experimenting, now.
Because, of course, COVID-19 is not the only crisis our generation faces. Indeed, as disruptive and as tragic as this pandemic is, it may only be an early warning shot from the future. In the coming decades, our governments are going to have to respond to more large-scale crises; some sudden, some gradual. They will need to embark on massive programmes to build new infrastructure, to upgrade our cities and homes, to build resilient, digital government services, to transform our food supply. Governments will need to be able to act rapidly and effectively to protect people’s wellbeing, to protect the environment shape new markets, and to build a resilient economy and society.
And it can all be done. Better, and for much, much less money than you might think. But only if we learn the lesson now, and fix procurement first.