Some more thoughts on land grabs and tricky statistics

I suppose you could label this post as my response to Ricardo and Marloes’s response to my post on their recent media brief on the correlation between governance and land grabs. First, I should say that this is all very exciting – it’s nice to have an actual debate about this. NGOs frequently ignore substantive criticism of their¬†analytical¬†work (to be fair, so do a lot of academics), so I must commend Ricardo and Marloes for their enthusiasm and willingness to get in touch and have a reasonable argument about all this.

I think we’ll likely to continue to disagree about when results should be presented (or at least¬†how they should be presented), so I’ll turn my attention to their three main technical points:

 

1) It’s not realistic to assume that investors target poor countries

True, but poor countries themselves might be more likely to put land up for sale. Discerning the difference between targeting and supply-side effects will always be difficult because we only observe actual land deals (in essence, the quantity `consumed’). But this is beside the point – spend fifteen minutes in an economics seminar and you’ll learn that a common way of challenging identifying assumptions is to come up with an equally-credible alternate story. I’ve shown that, at least in this very basic setup, income is a better predictor of a country having a land deal than governance. While my alternate story might be considered implausible (even if it does fit the data better), I really only put it up to point out how equally-flimsy the assumption of investor targeting is.

 

2) My last table is badly specified and then I forget to estimate a hurdle model.

Before delving into the technicalities of this argument, let’s briefly talk about burden of proof. It is Oxfam’s job here to convince us all that investors are targeting countries with poor governance, or at least that there is some consistent correlation between the two. By this very basic metric, I assert that the current analysis falls short, as it doesn’t provide enough evidence to reject the null of no relationship. One doesn’t always need to present and prove an alternate hypothesis, complete with fancy, well-specified econometrics, in order to disprove the one being asserted.

As far as the specification of the first two columns in Table 4 – sure, this is pretty much atheoretic wandering. I’m not going to assert that I’m cleanly identifying any individual channels, but seeing if Oxfam’s relationship stands up (I’m actually trying to help you here guys) once we start controlling for all these things. Multicollinearity doesn’t seem to be preventing some results from shining through. But yes, this is playtime with Stata, although I admit as much up front. See my point about burden of proof here.

Their final point is a technical one – I’m interested in whether, conditional on a country selling any land, governance is correlated with the number of land deals. Technically, this specification is subject to a form of bias due to selection on unobservables: for example, if hotter countries are more likely to sell land, and there is a correlation between temperature and the governance indicators, then estimates in columns (3) and (4) of Table 4 will be biased.¬† [OK this is not what their point was – see Paul’s comment below.]¬†Ricardo and Marloes would be happier if I estimated a model which took this selection into account.

But the problem is: as I point out at the end of my piece, I don’t really buy the selection equation in the first place, and this factors into their third point:

 

3). They take issue with my worry about “bias” in how land deals are reported.¬†

I’m worried the Land Matrix is a better measure of “number of reports on land deals” than “number of land deals” and that the measure of “have there been any substantial land deals” in the past ten years is really just a measure of “has anyone bothered to submit a news report to the land matrix on your country in the past ten years.” Ricardo and Marloes make a purely theoretical argument that reporting in the UK should be better than reporting in developing countries. If we were talking about general media reporting, I would be inclined to agree, but I’d be surprised if anyone is scanning British newspapers for land deals and submitting the data to the Land matrix.

Furthermore, consider the ¬†final hurdle a land deal must clear to get into the Land Matrix: “entail the conversion of land from local community use or important ecosystem service provision to commercial production.” This seems like it should only be possible in societies where a significant percentage of the population is involved in agriculture and where large scale commercialisation is yet to happen. Sure, the quality of the British government is one of the reasons there aren’t many dodgy deals going on, but we have to remember that Great Britain has already gone through the long process of moving from smallholder farming to relatively large-scale commercial production. Yet Ricardo and Marloes want to code Great Britain ¬†a zero and include it in the selection equation – I’m just not convinced.

How do we move forward? I’m happy that both Marloes and Ricardo want to continue working on this. This is definitely the best outcome – I can think of several ways that one could try and take it a little bit further

  • Let’s start exploiting the time dimension: we have a panel – let’s use it – although I do have a fear that, as several have pointed out, there won’t be enough meaningful variation in the WDI indicators across time to actually identify anything.
  • Number of deals and size¬†– these are, save for my kitchen sink regressions, currently unexploited. As is information on whether or not deals are international or national.
  • Let’s get more data!¬†I’m hesitant to throw a stake into the ground and say it’s time to make a call, especially when the data is as limited as it is. If we could get our hands on district level data (or, in my wildest dreams, GIS data) on land deals, we could start to say so much ¬†more about what’s going on.

Finally, a word to idle academics out there – I implore you to pay more attention to this stuff. We have a hard enough time encouraging replication of our own studies, but I think the world would be a much better place if we sat down from time to time and just tried to recreate “killer facts” that otherwise dominate the discourse. I didn’t start my analysis with any intention to go after Oxfam’s results, but it only took a little while with the data before I realised that the story was much more complex, and worth a second look.

Again, thanks to Ricardo and Marloes for a fun debate (I’ve offered them a second reply if they’d like).

 

Response from Oxfam: Governance, land grabs and tricky statistics

by Ricardo Fuentes-Nieva and Marloes Nicholls

It is encouraging to read the post from Aid Thoughts. We appreciate the time he put into Oxfam’s analysis on large scale land deals. Indeed, we were hoping that our blog would spark debate and bring more attention to this topic.

As a quick summary, we took two databases, the Land Matrix and the World Governance Indicators and found that land deals are more likely to occur in countries with lower levels across different governance indicators.¬† We specified that ‚ÄúThis analysis is only the first step towards a more in depth research project. Next steps include a more in depth analysis on the determinants of the number and location of deals‚ÄĚ

Aid Thoughts seems to take issue with the use of this kind of analysis when they are so preliminary. There are two things to say to this:  Firstly, and as Aid Thoughts acknowledges, there is other evidence in the development literature that points to the fact that land deals are concentrated in poorly governed countries. Our conclusions were not based only on our analysis and we used the best evidence at hand (both internal and external) to generate a better understanding of the problem (which Aid Thoughts actually helped with his critical review). So, we stand by our decision to publish the preliminary results.

Now, there are a couple of things to discuss on the technical front of his critique. Here are some:

 

1) Investors or governments?

AidThoughts replaces governance indicators with income per capita because they better explain the existence of land deals. This leads him to suggest that “Maybe investors aim for countries who are more willing to sell off land, not because they are poorly governed, but just because they are poor.” This is an interesting idea but if we put aside the regression tables and reflect for a moment, is it sensible to think that land investors are attracted to countries for being poor? Why would investors be attracted to the characteristics of poverty, such as poor infrastructure, limited public services and low levels of education and health? A more interesting hypothesis that AidThoughts raises, and which we think is worth exploring too, is that it might not ¬†be investors who target countries, but bad governments who sell the land of their citizens.

 

2) Truncated sample bias.

AidThoughts recognizes that running OLS with two control variables, as reported in Table 3, is not serious analysis (and yet he managed to muddle the significance of the estimators in his table). But what’s really puzzling is that, in order to prove his point, he then goes on to throw the entire kitchen sink of governance indicators into the next table (Table 4). These indicators are highly correlated amongst them, and it is difficult to find a sensible explanation to specify the model that way.

He then goes on to say of this table:

“In column (1), prior controlling for income, only one of the relationships we expected to see has returned: countries rated low on the rule of law index are more likely to have land deals. Political stability/violence is also associated with land deals, but unfortunately that wasn‚Äôt part of Oxfam‚Äôs theoretical model. Now, voice and accountability is positively correlated with land deals! Of course, most of these relationships vanish when we toss in income, although it is worth noting that the rule of law measure keeps its significance and sign. So the relationship between governance and land sales seems to be a lot more complex than the Oxfam brief is suggesting.”

That’s a lot of explanation for a badly specified model that includes highly correlated regressors. But that’s not even the most puzzling part of that table. AidThoughts then tries to explain the number of land deals with the same variables but he does not correct for the truncated sample (look how his sample drops from 212 and 183 in the first two columns to just over 50 in the last two). Ignoring the bias in the observed sample is a mistake and something we had identified as a problem, and that’s why we suggested exploring ¬†a double hurdle estimation to understand the issue better.

 

3) Reported land deals bias.

Aid Thoughts briefly mentions the potential problem of bias in the Land Matrix, but we don’t agree that he identified the right direction of bias. He argues that land deals are more likely to be reported in developing countries by diligent activists than in developed countries like the UK. On the contrary, we argue that land deals are much less likely to be ignored in richer countries with freer press, more access to information and better organized civil societies. Does Aid Thoughts seriously believe that a land deal can be more easily concealed in the UK than in the DRC?

Overall, we are very encouraged by Aid Thoughts‚Äô response. He mentions that he can be convinced of the problem with more data and more, better data is on its way according to conversations we’ve had with the people managing the Land Matrix.¬†So here’s our proposal for Matt: let’s work together – rigorously and objectively – on this issue in the next few months to try to better understand what’s driving the land rush. The problem deserves as much attention as we can give to it.

Land grabbing: whatever you do, don’t mention the G-word.

Large scale land purchases get some more media attention, this time from Jonathan Glennie over at The Guardian:

“A new¬†report on land acquisition¬†by the Munden Project/Rights and Resources Institute brings an important angle to the land “grab” debate. Rather than focusing on the ethics of land grabbing, the report makes the business case for working with local communities, arguing that failure to inform or fairly compensate affected locals heightens the risks to investors. Why? Because affected communities start to make life difficult for abusive or lazy companies, leading to massive unexpected costs or even an eventual full-scale retreat.

What is slightly disconcerting is that Glennie managed to write an entire article on land grabs while only using the world “government” once. NGOs and the media have largely painted the land grabbing story as a situation where evil companies are parachuting in and snatching land away (for example,¬†check out¬†Oxfam’s recent campaigning). In reality, land acquisitions which circumvent local property rights are only possible when governments themselves are incompetent, corrupt or overly-impatient. Of course campaigners realise this, but it’s much easier to set this up as story of evil capitalism than it is of governance, the latter being harder to sell and even harder to treat. I’m not trying to pick on Glennie for leaving out a lengthy discussion of governance in his article, but it would be nice for people to start using the g-word a bit more.

Governance, land grabs and tricky statistics

there-will-be-blood

“I’m a family man- I run a family business. And I heard your town scored low on the World Bank’s governance indicators.”

The last decade has been marked by a sudden increase in large scale land purchases in developing countries, a `land rush’ which has purportedly¬†been¬†driven by concerns over food security, food prices and a growing market for biofuels. The speed, size and lack of transparency over many of these deals, as well as their implications for the welfare and food security of those already living on the land, has led many to dub these large scale purchases as “land grabs.” This is a rather loaded term, but has successfully (and unfortunately) framed the context as one where anonymous, uncaring investors are systematically snatching land away from the poor and needy.

News reports suggest that at least some of this is happening – following the excellent Let’s Talk Land Tanzania¬†for just a few days reveals how¬†problematic some of these purchases have been. Yet, despite the ruckus these deals are creating, we still know precious little about their size and scale, the motivations and expectations of investors, the welfare impact on those in “grabbed” countries and the welfare impact on those in “grabbing” countries.

This lack of knowledge should be alarming rather than disarming, but while this is the perfect time for careful, dispassionate analysis and data collection, many have chosen to instead reinforce the simple “good” vs “evil” story I highlighted above. Take, for instance, this media briefing which Oxfam released last week, based on preliminary research on the relationship between country governance and land deals.

The two Oxfam researchers, Ricardo Fuentes-Nieva and Marloes Nicholls, use data from the World Bank’s Worldwide Governance Indicators (WGI) and the Land Matrix, which gathers data on media reports of land deals, to show that countries that had any land deals between the years 2000 and 2011 had significantly lower WGI scores than those that hadn’t had any. Here is the figure which they use to make their case:

land-grabs-and-governance

Again, this figure reveals that, across the four governance indicators considered by Fuentes-Neiva and Nicholls, countries with land deals consistently score worse than those without. How does Oxfam interpret these results?

Oxfam believes that investors actively target countries with weak governance in order to maximise profits and minimise red tape. Weak governance might enable this because it helps investors to sidestep costly and time-consuming rules and regulations, which, for example, might require them to consult with affected communities. Furthermore in countries where people are denied a voice, where business regulations are weak or non-existent, or where corruption is out of control it might be easier for investors to design the rules of the game to suit themselves.

So we have a pretty clear story here, right? Well, maybe not. Let me give a bit more structure to the above results by showing them as a series of bivariate regressions of the probability of observing a land deal in any given country between 2000 and 2011 and the average governance indicators for this period (the same data used in the Oxfam briefing).

oxfam1

Each column shows the results from regressing the probability of the country having at least one land deal during this period on each measure separately: voice and accountability, regulatory quality, rule of law and corruption (note that higher is `better’ for each of these measures). So far so good: in isolation, each of these variables is significantly* and negatively correlated with the probability of a land deal (i.e. countries that score poorly on each of these indicators individually are more likely to sell off land).

Yet, it’s a little strange that each of these seems to have about the same magnitude of an effect. We might expect some indicators to matter more. Also, for some reason, the Oxfam brief has left out two other WDI measures: political violence/stability and government effectiveness. Here is what the authors say about this exclusion:

Two of the Worldwide Governance Indicators Рpolitical stability and the absence of violence and government effectiveness were excluded from the analysis since there is no evident mechanism that would lead these aspects of governance to improve prospects for  investors.

OK – so we have a somewhat solid theoretical reason for excluding these variables. Presumably, we should not see the same negative correlation between these two WDI measures and land grabs. Table 2 below includes two extra columns in which I re-run the above results, but including both of the excluded indicators (PS and GE).

Continue reading