The International Land Coalition (ILC), in collaboration with prominent international developmental organizations such as the GIZ and CIRAD, recently launched its much-awaited global dataset of land deals, the Land Matrix. During last year’s conference on Global Land Grabbing in Sussex and with the publication of the ‘biggest study on large land deals’, the ILC has been enticing us with snippets of aggregated data on global land deals. It looked promising and I was eagerly awaiting to see its dataset to get a sense of its data quality. Besides, with such a large network of respected collaborators, I was confident that something more elaborate and scientifically rigorous would finally come on offer. It was becoming tiring to see the same deeply flawed figures being relentlessly rehashed (even by fellow academics); such as those, for example, of the Global Land Project, IFPRI, and the World Bank. The IIED wrote a succinct brief highlighting some of the problems with reports such as these.
The moment I read the Guardian’s announcement of the launch of the dataset, I was struck by a horrendous sense of dismay. In its investor top 10, I immediately noticed the entry of ZTE International in the DRC (2.8 million ha), Wuhan Kaidi in Zambia (2 million ha), and Daewoo in Madagascar (1.6 million ha). How can these be included? These land deals have never come to fruition. This is fairly common knowledge amongst those acquainted with such developments. ZTE only received 100,000 ha, which it is failing to develop; Wuhan Kaidi received only 79,300 ha; and Daewoo never received any land – a final leasehold contract for the land was in fact never signed (as the GIZ – one of the Land Matrix partners – already reported). The initial Daewoo MoU was also for 1.3 million ha, not 1.6 million ha.
I then turned to the actual dataset and my dismay grew even greater. While we were promised verified data, no thorough verification have appeared to taken place in practice. For example, for Ethiopia, a country I am more familiar with than any other, much of the data appears to be based solely on a report conducted by MELCA on biofuels. However, the data from this report is based entirely on a government dataset that records what areas of land prospective investors are interested in, not acquired. Also, it claims that the Indian government acquired 1 million ha. The links to data sources are broken and the only link that does work takes us to the Global Land Project (!).
The data for the DRC is even more embarrassing. It, for example, contains two entries based on an NGO presentation. The 3 million ha deal referred to be the presenter is the incorrect ZTE deal, which was already included in the dataset, and the entry on 2 million ha by Israel was merely an expression of interest. No such deal ever materialized.
On top of that, across the entire dataset there are numerous duplicate entries - rather sloppy. The 2 Karuturi land deals were, for example, referred to 5 times, the one Ruchi deal 3 times, and the SEKAB deal 5 times.
I pulled out 127 entries 0f the 466 Africa entries that are either duplicates, unverifiable, or simply incorrect. See Land Matrix data errors.
So, from removing these duplicates and inaccurate entries, the 70.2 million ha ILC claims has been acquired is reduced by 20.64 million ha to 49.56 million ha - and this is only from removing some Africa entries. This reduces the total area for sub-Saharan Africa (SSA) from 34.27 million ha to 13.63 million ha. So, based on this, it cannot be claimed that SSA is a major recipient of land investments.
While the ILC claims that all entries were circulated to in-country stakeholders for fact-checking, in reality it appears that this did little to improve data quality. This data is no better than the flawed reports cited above. The sad reality will be that given the extensive public promotion of the dataset, the media and other commentators will once again readily adopt these figures as fact without questioning their integrity.
For more information, contact: firstname.lastname@example.org