In a previous post, I extolled the virtues of Big Data in the fight against corruption, including in the important realm of government procurement. From the UK to Georgia to the Czech Republic, government procurement agencies have been collaborating with civil society groups to analyze their data, uncovering inefficiencies that range from the mundane to the outright corrupt. Governments are not alone: international development agencies like the World Bank are embarking on similar projects.
But there’s a problem. Big Data needs lots of data to work, entailing a high degree of government transparency and massive disclosures — sometimes called Open Government — that are sometimes at odds with the goals of anticorruption. In the case of government procurement, public data watchers need to know which firms bid for the project, at what price, and who won on what terms before they can play a useful watchdog role. However, as Rick has pointed out on this blog, public disclosure rules in procurement has the perverse effect of enabling private collusion. Cartels of contractors can agree amongst themselves to inflate their prices and select which among them will receive the contract, and are able to enforce their shady agreement because, of course, all offers are public.
Rick’s concerns seem to be directly implicated by the newly-proposed Open Contracting Data Standard, a push to “enhance and promote disclosure and participation in public contracting.” The project essentially asks every procurement agency in the world to upload their contracting documents onto the internet in a standardized manner that would encourage public oversight, including through the use of Big Data tools. So, is the push for open government procurement data doomed to backfire, creating collusion where perhaps it did not even exist before? Fortunately not. The increased risk of collusion is completely outweighed by the potential for the use of Big Data and other civil society monitoring techniques.
Let’s start by unpacking the problem. Government procurement processes are often riddled with corruption, and public disclosure of the bidding process makes it much more difficult for government officials to abuse their position. Specifically, procurement disclosure rules are designed to combat corruption in which officials are paid a kickback by a particular bidder in exchange for awarding them the contract; when the offers are public, the official has no choice but to select the best and cheapest bid. What incenses Rick, however, is the fact that these rules inadvertently facilitate private cartels by the very same transparency. Cartels of bidders can collude, agreeing amongst themselves to inflate their prices and even which among them will receive the contract, and are able to enforce their shady agreement because, of course, all offers are public. Thus, while sunshine has disinfected public procurement agencies, private cartels continue to extract rent and corrupt the marketplace.
Rick’s concerns have been addressed before on this blog. Daniel Holman makes the point that, even if transparency in public procurement just shifts corrupt activity into private cartels, this may actually be a preferable state of affairs. For one, if all of the corruption has moved to the private sector, enforcers will know exactly where to look to find violations. Another plus in Daniel’s eyes is the fact that, since the government’s hands are clean in the case of private cartels, it may have fewer qualms about exposing and prosecuting this corruption as opposed to when one of its own officials was bribed.
I agree with Daniel’s points, although I would like to add that it’s not clear to me that the conditions Rick describes are always sufficient to encourage collusion. Cartels tend to form in tight, closely-knit markets in which informal chains of communication between companies are more common — the cartels have a way of finding out if their members are playing by the rules whether the contract information is publicly available or not. Thus, less transparent procurement systems may have the effect of inhibiting public scrutiny without inhibiting cartels.
But I admit that Rick is right overall — full transparency of the procurement process does at least somewhat increase the risk of private cartels forming. My argument, however, is that this increased risk is completely mitigated by the potential for civil society monitoring using Big Data and other techniques. Properly implemented, Big Data should be able to tell citizens and governments where collusive inefficiencies lie. Under the harsh glare of public monitoring, amplified by the power of modern data analysis techniques, it’s unlikely that private cartels would get away with much rent-seeking activity for long.
Proper implementation is key, however. Governments should not release information in such a way that civil society groups cannot easily engage with it — or else collusive actors might go undetected and unchallenged. The Open Contracting Data Standard presents the public with opportunities to combat collusion that overwhelm the risks only because it would streamline and standardize government disclosures in such a way that Big Data techniques could be easily applied to the problem of detecting inefficient rent-seeking behavior. OCDS eases the process of discovering rents by allowing anyone to compare similar contracts across jurisdictions and markets. With this enhanced ability to discover market prices, the public can play a larger role in eliminating all sorts of inefficiencies (whether they are caused by corruption or, more commonly, simple waste). The power of this capability increases at pace with the number and quality of contracts in the database.
Here are some specific examples of open data systems and Big Data techniques being used by civil society to detect inefficiencies, including cartels:
- Transparency International’s chapter in Georgia analysed 430,000 government purchases made available through that country’s new open data initiative and, by cross referencing with other sources of open data, discovered an inappropriate correlation between benefits received by public officials and contracts awarded to private parties. Uncompetitive procurement processes awarded $150 million in contracts directly to companies owned by members of parliament and public officials. Companies not owned by public officials that also received offers without competitive tender were responsible for 60% of disclosed donations to the ruling party, with the average donation about 4% of the contract value.
- TI Slovakia used its country’s new open data system to expose serious inefficiencies in the field of hospital procurement. Some inefficiencies were so gross as to be near-certain indications of corruption, such as buying CT scanners on two different occasions within the same year but at prices that varied by as much as 100%. More than half of all hospital tenders were uncompetitive, with only one bidder, and those that did attract other bidders yielded contract prices 20% cheaper for the same services.
- The World Bank’s Road Cost Knowledge System (ROCKS) uses data from Bank-funded roads projects to shed light on a procurement area that is fraught with cost overrruns. In India, for example, almost half of all roads projects overrun their bidding price by more than 25%, and the average roads project worldwide sees similar overruns of 20%. ROCKS helps procurement agencies and the citizens who watch over them gain a better sense of what they should be paying for a given stretch of road.
These efforts, culled from a report by the Center for Global Development, are good examples of what civil society can do for their part in the procurement process. Citizens, who are understandably interested in public procurement (as they are paying for it), can investigate potential inefficiencies and corruption only if they have access to open bidding information. The Open Contracting Data Standard would facilitate this kind of citizen monitoring, making corruption-detection in government procurement a world-wide effort — if one mile of road is more expensive to procure in one country than a comparable stretch in another, citizens with OCDS data in hand can take steps to understanding if the price difference is the result of a rent-seeking cartel or something more benign.
Again, I concede that some types of transparency are not desirable solutions to corruption. Rick is right that some procurement markets would not benefit from increased transparency, and perhaps for these markets the OECD’s Guidelines For Fighting Bid-Rigging will be especially useful. However, for what I feel are the majority of markets, the public has much more to gain from procurement transparency than do cartels. Most procurement processes are already designed to be transparent, but many existing platforms for public engagement are abysmal — the only entities that end up using the open bidding information are, as Rick suggests, the cartels. Initiatives like OCDS could invogorate civil society’s role in cleaning up procurement, and the anticorruption community should support it. It is one thing to collude in a tight local market with limited information, but collusion is much harder when the marketplace is worldwide and transparent.
some additional civil society and government initiatives using big data in public procurement to detect corruption:
-Czech Republic: http://www.zindex.cz/