Strong Evidence that Donors Use Development Assistance to (Try to) Influence Elections

Researchers have scrutinized foreign aid’s effects on poverty and growth, but anecdotal evidence suggests that donors often use aid for other ends. We test whether donors use bilateral aid to influence elections in developing countries. We find that recipient country administrations closely aligned with a donor receive more aid during election years, while those less aligned receive less. Consistent with our interpretation, this effect holds only in competitive elections, is absent in U.S. aid flows to non-government entities, and is driven by bilateral alignment rather than incumbent characteristics.

That’s the abstract from an important new paper by UC-San Diego economists Michael Faye and Paul Niehaus, forthcoming in American Economic Review. Technically, official development aid (ODA) is supposed to be about promoting economic development and improving popular welfare. Nevertheless, Faye and Niehaus show a strong link between election cycles and aid flows that fits what we would expect if aid were also being used for political ends. In cases where elections are competitive, donors crank up the aid to friendly governments facing tough elections while reducing aid to hostile ones. In cases where elections aren’t competitive, aid flows don’t vary much around elections (why bother, right?). Meanwhile, assistance to non-governmental organizations and opposition groups from the U.S.’s National Endowment for Democracy (NED) follows the same cycles, but the pattern is reversed (albeit not statistically significant): assistance to opposition groups goes down around election time in countries with friendlier governments, and it goes up around election time in countries with more hostile governments.

All in all, it’s a pretty compelling set of results that should put another big dent in the “development aid isn’t political” narrative.

Thanks to NYU’s Cyrus Samii for pointing this paper out on Twitter.

Leave a comment

6 Comments

  1. Jack

     /  January 25, 2012

    I should probably read the article and look at the data, but do the authors account for the possibilities that (1) development assistance goes to more responsive regimes (presumed to use it for development ends) and (2) election time puts countries back on the map in the eyes of many governance reformers?

    Reply
    • Jack, I think the design addresses those concerns by 1) using fixed effects to focus on change in ODA flows over time within countries (so it’s not about cross-country differences in governance) and 2a) finding that the direction of this variation over time is correlated with a country’s policy affinity with the U.S. and is conditional on the election’s competitiveness (so it’s clearly not about election time as such). But I’m not the author, of course, so you really should read the original and judge for yourself.

      Reply
  2. Antonio

     /  January 25, 2012

    The whole paper without a graph of the data! how can I believe it? Nice blog here!

    Reply
  3. Andrew

     /  January 26, 2012

    The analysis is only as good as the data it is based on, and there are serious issues with the reliability and validity of the data used in this study. The USG data on DAC in particular is quite problematic, and for most of the 1990s should really just not be used. The NED data needs to be cleaned quite a bit to reflect activities accurately, and and the bulk of election assistance from the UK and Germany went through the political foundations, not the aid departments. The selection of donors is also an issue, as it misses donors that do more election work than the ones selected, plus UNDP has been the channel of choice for some major donors.

    Another demonstration of the perils of using datasets without a better understanding of their source and nature.

    Reply
    • Thanks for pointing that out, Andrew. You probably know the DAC data better than anyone else, so that’s an important reminder. Given what you know, do you think there’s any chance the inaccuracies are the source of the pattern the authors observe? And would you expect those omitted flows to be distributed less politically than the DAC money, or more?

      Reply
      • Andrew

         /  January 27, 2012

        I need to think about that a bit, and I’m jammed up with some deadlines at the moment…first, more on why the data should be viewed as flawed.

        Look, the DAC/CRS data are problematic for a number of reasons, most importantly because of coding issues, ie what we should recognized as inter-coder reliability. As with any massive data gathering exercise (system of national accts, etc.), there’s a large guidance document on how to code. That said, there’s no real verification of coding application at OECD, the donors don’t always follow them very well, and the category definitions are not always clear. One example related to elections is the common & important activity of voter education, which is typically implemented through local NGOs and/or media. One donor might code this as “Elections,” another as “Civil Society,” another as “Support to Local/Regional NGOs,” and another perhaps as “Free Flow of Info.” If the assistance worked primarily with the electoral management body (EMB, often called central election commission), the donor might code that as “Elections” or even the generic category of “Government Administration.”

        Coding reliability problems exist for a donor from year to year, too. For the 50+k record dataset I put together on USAID funding (1990-2006) that was used by Finkel et al, I spent months and months researching projects and re-coding things to make it all as accurate as possible.

        Another coding issue is that donors like USAID may use multiple categories for an activity. For example, an election assistance project that strengthens the capacity of the EMB, updates the voter registry, supplies all of the polling place materials, undertakes a voter education campaign, and develops domestic election monitoring by local NGOs could have parts of the project coded under “Government Administration,” “Elections,” “Civil Society,” “Free Flow of Info,” “Human Rights,” “Statistical Capacity Building” under Other Social Infrastructure & Services, perhaps the “civilian peacebuilding” under Conflict Prevention, and perhaps something under Support to NGOs.

        The assistance approach taken by a donor matters, too, as some donors prefer working with state institutions (eg France) and others through their own country’s and/or the target country’s NGOs (eg Sweden), so that in either case it is quite possible that the donor coding will not reflect the actual activity. As I mentioned in my original comment, UNDP became the favored organization for European donors, if only in some cases because funds could be pooled. Taking USAID as a similar example, its assistance may have gone through its regional bureaus to such regional organizations as the OAS or the AU, or perhaps through one of the Washington-based procurement mechanisms. Oh, also, it has only been the last couple years that the data for State have been any good, in fact they didn’t report at all for years.

        USAID data was also horribly misleading/inaccurate for years, although the final roll-up numbers tended to be correct. One reason for this is that the first few years of assistance to post-communist countries actually occurred under and was reported at the regional level. UNDP does not provide data, period. I was unable to get data from the German stiftungen.

        When it comes to foreign assistance data, particularly democracy/governance data, DAC/CRS is not the way to go (I’m not wild about the authors’ political alliance variable, either, BTW, but that’s for conceptual reasons). This is why I did some collaborative work with PLAID, now called AidData, which was focusing on the project level. They were/are cleaning up and re-coding data in useful ways, but USAID and UNDP are still stumbling blocks – USAID because its prior reporting is viewed as final and official, and UNDP because they kept poor records and have no interest in transparency.

Leave a Comment

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

Follow

Get every new post delivered to your Inbox.

Join 7,745 other followers

%d bloggers like this: