Wednesday, 18 February 2015

The Customs Computer $30 Million Blowout

Guest post by Daniel Silva from the Importer’s Institute

Yet another big government IT project has gone off the rails. This time it is something called the Joint Border Management System (JBMS), a system designed to upgrade and link systems currently operated by Customs and the Ministry of Primary Industries (MPI). It was supposed to be finished by the end of 2012 at a cost of $75.9 million. Three years later, the development is still not complete and the budget has ballooned to $104.1 million.

This is merely the latest in a series of similar debacles. They are practically unavoidable, given the model used by government for IT procurement. It has been described as the “waterfall” model. Mike Bracken of the UK’s Government Digital Services described it as “writing most when you know least.” A group of public servants write massive tender documents attempting to guess the needs of end-users years in advance. The scope of the projects is such that only a few large consultancies can qualify to tender. The selected consultancy then gets on with over-running budgets and missing delivery deadlines.

When it all becomes public, everyone runs for cover and the hapless Minister nominally in charge is left to take the flak. In the JBMS project, Customs, MPI, the Department of the Prime Minister and Cabinet and the State Services Commission were all called to put out the fire. Official Information requests were declined (on the usual grounds of ‘commercial confidentiality’) and an embarrassing law suit with IBM was avoided, but not before a few more millions were spent getting legal advice from Crown Law and Chapman Tripp. Those suits don’t come cheap.

Does it have to be this way? After all, banks do on-line banking, Amazon sells books, clothing retailers of all sizes sell clothing, Trademe manages an auction site and airlines sell tickets on-line. None of these well-functioning systems were developed using the waterfall model. They were not the product of any grand design and all started as relatively small projects, run by in-house teams or small-scale contractors. Unlike public servants, the owners of those businesses did not have access to lavish amounts of other people’s money. Many of the early versions of these successful applications failed miserably, but were soon replaced by others that did not.

Matt Ridley, a member of the British House of Lords, said in an article published in The Times,

[The systems that] succeed allow for plenty of low-cost trial and error and incremental change. It’s the mechanism Charles Darwin discovered that Mother Nature uses. Rather than a grand ‘creationist’ plan or a big leap, natural selection incrementally discovers success through trial and failure. From the English language to an airliner, everything successful has emerged by small steps.

The current Customs system, Cusmod, was developed around the same time as trackstock, the Warehouse Management System that we use at DSL Logistics. We like to say that trackstock is very good indeed, because it incorporates fifteen years of mistakes. Every time something goes wrong, our small in-house team of developers comes up with solutions which improve the functionality of the product. Cusmod, on the other hand, looks substantially the same that it did fifteen years ago. Of course it is out-dated and no longer fit for purpose. Instead of breaking the system up into smaller units and developing the capability to maintain them in-house or by using local developers, the departments went for another big bang approach and propose to replace the whole thing by spending over $100 million of taxpayers money.

The idea that a group of public servants can specify the myriad requirements of users, years in advance, is nothing short of arrogance. Yet, New Zealand is awash with very smart developers that deliver solutions to businesses of all sizes, day in and day out. Unlike massive projects like Novopay or JBMS, those systems actually work, in a way that no public sector IT system has ever done.

The rationale for JBMS was never very clear. At one stage, it was trumpeted as making it easier for businesses to register with the bureaucracies, a step that every business must undertake exactly once in its lifetime. The need for a ‘single window’ arose from the decision to keep two separate agencies working on border control, contrary to what every commission of inquiry convened on that topic over the years recommended. A single border control agency is the norm in comparable countries. It soon became apparent that the real purpose of the exercise was to upgrade the control systems of both agencies, while keeping two chief executives and their coteries of deputies.

Customs already records details of every transaction in a modern relational database. That provides the mechanism for the collection of statistics and the management of the revenue collection functions. The same data can also be used for intelligence purposes. There is no discernible need to create a new mega system to improve intelligence gathering. If the data being collected is not sufficient, new fields need to be mandated. Customs officers need to receive better training on how to query their databases and produce better analytics. If the tools that they have at their disposal are not good enough, then any of the many local software developers can provide them with query/reporting tools by lunch-time, at a fraction of the IBM costs.

Does such an approach work? It does. According to Matt Ridley, British minister for the Cabinet Office Francis Maude

began by centralising controls so that he had to sign off any IT contracts of more than one million pounds (now raised to five million), then built up an in-house capability to offer cheaper and better design, and opened procurement to smaller companies. Government contracts with outside IT suppliers are now shorter and smaller. Some of the savings on offer were so vast that civil servants refused to believe them. In one case, 98.5 per cent of the cost of an existing contract was saved by letting a contract to a small British business rather than an incumbent multinational IT firm, and it worked better.

Internal Affairs Minister Peter Dunne represented New Zealand at the inaugural meeting of the D5 Digital Leaders’ Summit in London in December 2014. The D5 is a grouping of five nations – United Kingdom, New Zealand, South Korea, Estonia and Israel – considered amongst the most advanced in the provision of on-line government services. Its establishment was a British government initiative. We hope that Mr Dunne learned something at that meeting.

Daniel Silva is the head of the Importers Institute, an informal national association of New Zealand importing companies keeping members informed on topical issues of interest, and representing importers’ interests before policy makers and the public.


  1. Now, imagine what will be the biggest govt IT project yet, upgrading the IRD system. Initial budget is for $1.5 billion (yes, I said BILLION), last I heard have so far spent $50 million (yes, MILLION) just trying to sort out the contractors, and that's not even completed yet :)

    (Be shame if the whole damn system fell over and the government could collect its booty.)

  2. Typo ... 'couldn't collect its booty.'

  3. Enjoyed this. He nails it & knows of what he speaks. Being in software, it is round upon round of small trial & error, and the smaller the team the better. Specs should be max 3 pages, ideally 1. Govt types find such too unnerving, and so pay alot for control & ass-covering paperwork.

  4. @Mark, IRD system -- someone's pulling figures outta their ass just to set the tone. For all the buy NZ made pap, that won't get built here. Maybe a few pieces. It will move veeery slowly.

  5. I really like your post.It's very informative and interesting. I really appreciate that.thanks for sharing information with us.


1. Commenters are welcome and invited.
2. All comments are moderated. Off-topic grandstanding, spam, and gibberish will be ignored. Tu quoque will be moderated.
3. Read the post before you comment. Challenge facts, but don't simply ignore them.
4. Use a name. If it's important enough to say, it's important enough to put a name to.
5. Above all: Act with honour. Say what you mean, and mean what you say.