This section reflects on the findings and outlines some recommendations for donors, recipient countries, software vendors, and the wider IATI community.

3.1 Donors

There are two main sets of recommendations for donors: firstly, to improve data quality, and secondly to prioritise support for similar initiatives in future. Donors are listed first in this section because addressing these two sets of challenges will provide a strong boost to efforts to use data as part of country systems and elsewhere.

3.1.1 Improve data quality

There were fifteen donors whose IATI data was good enough to begin importing into the AIMS. However, there were some donors whose data was not possible to use, and many donors’ data had some shortcomings. There should be a compelling business case for all donors to improve the quality of their data. Some small changes at headquarters could save donors hundreds of staff days per year, as well as making their aid more effective and reducing transactions costs on recipient country governments.

  1. Ensure you’re publishing projects. This may seem like a really basic point, but several donors were publishing activities that are not meaningful or useful units of aid. This issue came up to varying degrees with UNFPA, UNICEF and USAID data (see section 2.4.1 above).
  2. Use parts of the IATI Standard that can help avoid double-counting. Donors should refer to related activity identifiers, particularly where they are involved in projects with other donors. The documentation and guidance on the IATI Standard website should be improved to clarify how this should work (see section 2.5.2 above).
  3. Provide disaggregated financial data. Granular financial data is important for being able to slice the data in different ways and may have significant implications for currency conversion depending on the degree of fluctuation in exchange rates. This meant we couldn’t use Germany’s data, and could also be improved in the World Bank’s data (see section 2.4.3 above).
  4. Improve the identification of organisations. Where “implementing organisations” state generic categories of organisations, it is not possible to map to specific organisations in the AIMS. It is also much harder to avoid double-counting. This was an issue with DFID and UNDP data (see section 2.4.2 above).
  5. Provide frequent and timely data. Annual re-published CRS data is of no value in country systems as it is far too out of date to inform decision-making. This was an issue with Japan’s data. Quarterly data is the minimum that is requested from donors to enter into the AIMS, so donors that are publishing less frequently than this will need to continue manually entering financial data. Monthly data is also required in ERD for both budgeting and monitoring project execution.
  6. Provide forward spending data. Some donors are providing this data in the AIMS but it is still not available in many donors’ IATI data. This is vital data for budgeting processes.
  7. Ensure your data is being published in an appropriate language. In a couple of cases, there were titles and descriptions that were not published in English, even where it was stated that the data was in English. In countries where the use of the English language is not so widespread, publication in other local languages will become more important.
  8. Look at (and use) your own data. In a couple of cases, the data that is being published is hard to reason with. Using a tool like D-Portal to take a look at the data could help to bring to light some basic issues with the content and structure of the data.
  9. Use IATI data internally. If IATI data was used for internal processes, donors local staff would be much more familiar with and comfortable with using it to report to recipient countries. Any issues would also be much more quickly identified.

3.1.2 Prioritise support for (and fund) more work to use IATI data

This work was only made possible through the support of DFID. It was vital to have some time for methodological development and to allow some significant management, oversight and course correction over the period of software development. It also helped that, following a USAID outreach mission in September 2014, there was greater awareness among donors than might otherwise be expected. However, in order for this work is to be taken forward elsewhere, we recommend greater investment and outreach to donor country offices.

  1. Greater investment is required. More research, methodological development, technical support and software development is needed to take this work forward and to develop sustainable, scalable processes. Much of this work could have been undertaken two years ago had funding been available. There is a clear opportunity to create a virtuous circle here, whereby the use of IATI data in a few countries provides strong incentives to improve the quality of data and for others to begin publishing. Support could either be channelled through the IATI budget or donors could commit to directly support particular efforts in various countries, but some coordination would be helpful.
  2. Need for outreach to donor country offices. There is a need for greater sensitisation of donor country offices to the potential of IATI data. Country offices should be encouraged to view this as important work – not just in terms of transparency, but also improving the effectiveness and impact of their projects. At the same time, country offices need to recognise that efforts to begin using IATI data will take some time as techniques and software are developed. So there is also a need for patience, as well as a nuanced understanding that progress will depend on the quality of each donor’s data.

3.2 Recipient countries

The leadership role of the Government of Bangladesh was vital for completing the work in such a short period of time. Political commitment to IATI and a good understanding of some of the technical and policy questions raised in beginning to use this data created a supportive environment. The decision to proceed with a home-grown AIMS – giving the government control of the source code – and use of local developers provided vital flexibility. The government was also fortunate to have some in-house technical capacity early on which supported procurement.

  1. Push your vendors to incorporate IATI data in your AIMS. If you are procuring a new AIMS, ensure not just that it is “IATI compatible” but that it really handles things such as double counting and merging projects when they are published by multiple organisations. Consider in advance how importing IATI data would relate to and support your existing business processes. Consider hiring technical assistance to support the process of procurement and monitor the process of development.
  2. Take a pragmatic approach to using IATI data. IATI can be used to reduce the burden of data collection and improve quality for those donors that have good enough data. That will provide more time for focusing on donors whose data is difficult to get hold of, and for shifting to using the data to improve effectiveness and impact of development cooperation. Donors can continue to use manual data entry if the data is not good enough, or if certain important fields are missing.
  3. Take steps to prepare for IATI import. Sensitise donors at the country level to the potential of using IATI data. Write letters to your donors indicating your interest in beginning to use their data. Encourage donors to ensure they are using the correct project IDs in the AIMS – the fact that donors were already using the correct project IDs fairly consistently made it much easier to map IATI activities onto AIMS projects

3.3 Software vendors

Vendors have a key role to play – both as the main technical conduit to recipient countries already using AIMS, and in developing techniques and software to begin to handle this data.

  1. See if the findings from this work could be relevant in your software or if you could improve on them. Consider in particular techniques for avoiding double counting, delegating and merging projects, the ability to merge projects from IATI and AIMS in a somewhat nuanced way, and the ability to exclude individual project sub-components. The challenge is to develop an interface that is simple but also allows significant flexibility to allow import to be tweaked for each project and allow a mix of data from IATI and AIMS to ensure you can get the best data from each system.
  2. Focus on allowing humans to make decisions only they can make, but simplifying everything else they don’t need to think about. Things such as retrieving data, converting to a standard version of the IATI Standard and currency conversion should all be automatically handled behind the scenes. It is clear that the interface developed could be further simplified along these lines: there are probably some options which are exposed in the interface which could also be hidden (for example, where the impact of selecting different options may be limited, or where users consistently select a particular option).
  3. Consider this a core or standard feature of any AIMS going forward and a fundamental part of your business model. As it is now clear that IATI data can be used in country systems, AIMS will likely be placed at a competitive disadvantage in future if they are not able to handle IATI data. Investing in good quality IATI import at this stage – and not as an add-on or afterthought – will pay dividends in future.

3.4 IATI Community

The IATI community also has an important role to play in enabling and driving this work forward. Some of these are recommendations for the Members’ Assembly, Board or TAG to consider, but they are relevant for all.

3.4.1 Setting goals and vision

  1. Set an ambitious goal for use of IATI data. A statement of intent to meet certain targets could be helpful for providing direction and momentum. For example, the number of countries’ annual aid reports that use IATI data for at least 50% of their total aid within the next three years. This doesn’t necessarily mean that the IATI Secretariat needs to do the work to achieve that goal – indeed, that may be unrealistic given available resources. However, a goal for the IATI community (perhaps a collective commitment from the Members’ Assembly) would provide a focus and a benchmark to assess progress over time.

3.4.2 Data quality

  1. Encourage donors to use components of the IATI Standard that help to avoid double-counting in multi-donor projects. Much of the work around traceability to date has focused on relationships between donors and their NGO implementing partners. However, donors could potentially make some rapid improvements in data quality from working together to refer to each other’s activities in projects involving multiple donors. Major implementers such as UNDP and the World Bank should be two priority targets for such efforts.
  2. Focus efforts to improve data quality on official donors. There are a handful of donors that are still publishing very poor quality data. Of particular concern are cases where donors are not publishing projects as the main unit of aid, because the data is then either impossible or very difficult to use. Centrally-provided technical support should work with such donors to improve the quality of data they are publishing. There is likely to be a much higher return on investment from improving the data quality of a small number of organisations with large volumes of spending, than focusing on many small organisations. This should be combined with more intensive outreach to those large donors who are not publishing data to IATI, or only publishing historical CRS data. In Bangladesh, two of the four largest donors are Japan and the Islamic Development Bank.
  3. Develop mechanisms for feedback on data quality. There is no consistent way for sharing issues with individual donors’ data. Such a mechanism would be helpful for users of the data to understand the challenges they are likely to face when trying to import IATI data, as well as helping donors to track how their data is being used. It may also be interesting to consider how to share feedback where manual changes to IATI data were necessary at country level.
  4. Push more strongly for monthly data, and restate that quarterly data is the minimum required under the Standard. Where data is published less than quarterly, it is fundamentally not useful. Countries still need this information, so they won’t use IATI data – they will go back to donors with manual data requests.

3.4.3 IATI Standard

  1. Improve guidance on mapping together activities reported by different organisations to avoid double counting. A consistent methodology for stating that an activity is a contribution to a trust fund, the trust fund itself, or a project funded out of it would be helpful for automating more of the process of identifying potential double-counting.
  2. Organisation identifiers should be a priority area for improvement. The Standard should make clear that specific organisations should be identified rather than broad categories. DAC Channel Codes do provide some useful information in categorising organisations; we recommend incorporating these categories into the IATI Organisation Type codelist. Finally, developing a consistent methodology for organisation identifiers for public bodies would help to automate more of the process of importing data.
  3. Avoid disruptive and breaking changes to the Standard. We recommend a more careful consideration of potential costs and benefits of upgrades to the standard – which can be disruptive and difficult to deal with. Changes that are not backward compatible should be avoided wherever possible. They may have significant negative effects on the structure of systems already using the data and it may become conceptually challenging to merge data from different versions into the same system. We also recommend road-testing or piloting changes to the Standard to make sure they work with a range of donors’ data before formal agreement.
  4. Clarify the disbursement channel codelist. The Disbursement Channel codelist should help to identify whether funds are “on budget” according to a couple of different definitions. However, it appears that one of the codes is being misinterpreted. In the statement “Money is disbursed directly to the implementing institution…” it seems highly likely (given the other values on that codelist) that “implementing institution” refers to government institutions. However, this should be clarified as it is currently being used inconsistently.
  5. Allow exchange rates and interest rates to be specified on transactions. Exchange rates should be provided if the agreement stipulates a fixed rate that all parties will adhere to. Interest rates could also be provided in the case of loans. Both of these fields could remain voluntary, but they would be useful for debt management.
  6. Provide a place for publishers to state that their IATI data is their official data. An addition to the standard stating whether donors were happy to be held accountable against their IATI data would make it much more palatable for end users, and reduce some of the process and accountability issues around automatic import.

3.4.4 Infrastructure

  1. Improve the IATI Datastore. The IATI Datastore provides a good back-end for accessing original XML data, but could benefit from several improvements. Improvements to documentation and the user interface would make it easier for users to obtain extracts of the data. It should also provide automatic conversion to a requested version of the IATI Standard so that each system does not have to implement conversion manually each time. There should be an explicit commitment to converting data to the latest version of the Standard whenever an upgrade takes place. When organisation identifiers change, the Datastore should retain an up to date mapping between old and new identifiers and return all relevant data, as users will not necessarily know that the organisation identifier has changed. Clarity on support and maintenance that will be provided for the IATI Datastore going forward (in the style of a Service-level agreement) would also help users to make informed decisions about whether to use this platform and the extent to which they can rely on it.
  2. Improve D-Portal. There is a clear need for potential users of this data to take a look at the sort of information contained in the data they are considering to using. D-Portal is the obvious candidate for pointing people towards, but there are a few areas where improvements would help. Support for handling hierarchies of activities would improve the way that organisations like DFID’s data appears on the site. On the project view, showing more of the data included alongside a particular activity (for example, locations) would be helpful. It would be helpful to provide an overview or summary of data on individual project pages – for example, total disbursements and commitments per year, rather than having to look through the full list of transactions. It should also be possible to see the data in USD on the project page (this is only listed on the page listing all projects). It is also notable that D-Portal is not able to present a figure which aggregates the contributions of multiple reporters in a single recipient country, despite this being a very common request. Finally, the user interface could be improved by improving the way information in multiple languages is displayed (for example, look at Canada’s projects).