Does government funding silence Australian NGOs?

Man with a megaphone (

Should NGOs take money from the government? It’s one of those long-debated questions of development. There’s an obvious argument in favour: the money can be put to good use. And yet the counter argument is clear enough too: take money from the state and you give it leverage over you. If you do or say things it doesn’t like, it can cut your funding. At times you may directly be told what you can do or say. In other instances your own preemptive second-guessing may have the same effect. Or at least that’s the theory. But what about in practice?

The data we gathered as part of the NGO internet use content analysis we wrote about in our last blog offers us a test of this theory in the Australian case. From our content analysis we knew the extent to which each of the largest 50 Australian aid NGOs used different internet media to raise public awareness about development, and the extent which they used internet media to encourage people to take action on development-related issues. We were then able to combine these data with data on NGOs gathered by the Australian Council for International Development (ACFID) for the compilation of their 2015 annual report. In particular, we were able to obtain information on how large each NGO was, whether they were religious or not, and the proportion of their revenue that came from the Australian government.

With the combined data we were able to run regressions to see whether there was a relationship between the proportion of NGOs’ revenue that came from the Australian government aid program and the extent to which they used the internet to raise awareness and to encourage people to take action. We ran two regressions: one for “awareness” and one for “take action”. In each, the dependent variable was the average for the category over all of the different media types we studied. (So, for “take action”, it was take action for each NGO averaged across their front page, their get involved page, their Facebook feed and their Twitter feed.) In each regression we included two control variables (whether an NGO was religious or not, and the NGO’s size in terms of revenue), alongside the main independent variable of interest: proportion of revenue that came from the government. (Full details are in the methodology section of the discussion paper this blog post is based on.)

The result we were expecting was that more government funding would be associated with less internet media space devoted to raising awareness and encouraging people to take action. The two charts below show what we got. (These are the effects of government funding with the influence of the two control variables taken into account.)

Marginal effects - Awareness and Take Action

As the charts above show, we found the exact opposite of what we were expecting. If increased dependence on funding from the Australian aid program constrained NGOs’ ability to engage in awareness raising or encouraging people to take action on development issues, the lines on the charts would be downwards sloping, but they aren’t. In fact they are positive. The regression coefficients that the charts are drawn from are both statistically significant at p<0.1, which is reasonable for a sample of 50. And the findings performed pretty well in robustness tests. Ideally, we would have been able to add more control variables into the equation, but limited data and a small sample prevented this. So we are quite confident of the positive relationships, but are not absolutely certain. We can, however, be very confident that the relationships are not negative.

Of course, we do need to add some caveats. Showing that, on average, government funding is not negatively associated with awareness raising and encouraging people to take action is not the same as saying governments never use the leverage of funding to silence unruly NGOs. And although there is some tension at present between NGOs and the current government over issues like aid cuts, it’s possible that had we undertaken our study in a time when government/NGO relations were particularly oppositional, results might have been different. Also, our coding categories were general categories to do with “awareness” and “take action”; we might have found different answers if we focused only on activities that assertively targeted the government. What’s more, we’ve only looked at how NGOs use their internet presence. The relationship might plausibly have been different had we studied other media. Because of these caveats, we want to be emphatically clear: we’re not telling NGOs they can race out and take government funding confident this will not at some point constrain their ability to speak their minds. Practically speaking, that would be unwise.

However, thinking more theoretically, and even taking on board the caveats above, the positive relationship is an intriguing surprise. In our discussion paper we explain why we think it exists. In essence, our argument is that to a degree, for those NGOs that receive it, government funding reduces their need to obtain donations by every means necessary. And this in turn affords these NGOs the luxury of devoting some more of their internet media presence to fostering public engagement on development issues.

Terence Wood is a Research Fellow at the Development Policy Centre. Terence’s research interests include aid policy, the politics of aid, and governance in developing countries. Sachini Muller is a student at the Australian National University, and interned at the Development Policy Centre in Semester 1, 2016. Her interests lie in research and the aid and development sector, particularly in Asia and the Pacific.

This post is the second in a two-part series summarising Devpolicy Discussion Paper 47; find the first post here.

Terence Wood

Terence Wood is a Fellow at the Development Policy Centre. His research focuses on political governance in Western Melanesia, and Australian and New Zealand aid.

Sachini Muller

Sachini Muller was a Research Officer at the Development Policy Centre. She is currently completing a Master of Globalisation at ANU.


  • In practice, there are many complex factors that influence the decision of when and how NGOs will advocate around government policy – the proportion of government funding is clearly a factor but it’s only one of many.

    For example, Save the Children was founded by an outspoken advocate and our theory of change explicitly emphasises the importance of advocacy in achieving systemic change. Hence, advocacy is a core part of our DNA, or our culture as Patrick says. Internal capability to engage in advocacy effectively is another, particularly the sophisticated and politically nuanced advocacy required for a topic like refugee policy. Leadership, both management and Board is critical.

    Operational issues weigh heavily: will it place staff safety at risk? Will public advocacy undermine your ability to influence behind the scenes? In my experience, public advocacy should be like an iceberg – just the tip. Effective advocacy most frequently occurs behind closed doors and therefore can be extremely difficult to assess from the outside. And what is ultimately in the best interests of the vulnerable groups you are seeking to support? Calculating this requires a very difficult assessment of the implications of reduced impact from meeting people’s immediate humanitarian needs, with the potential benefits from systemic change should you be able to change government policy. As a child rights organisation, this balancing act was something we constantly assessed in relation to Nauru, to ensure we were meeting the Convention’s overriding obligation to act in “the best interests of children”.

    The broader regulatory and political environment is also relevant. Gag clauses, the Freeedom to Advocate Act, the Australian Border Force Act (for refugee issues) etc are only the most blunt tools – government funding decisions are normally much more subtle.

    This is an important debate and the above analysis is very thought provoking but given the complexity of the issues we are seeking to influence, I would be wary of drawing too many conclusions and instead encourage continued research and exploration!

    • Hi Paul,

      Thank you for an excellent comment. I think you’ve done a great job of explaining the complexity of the decisions all NGOs must go through when deciding about advocacy, and when deciding about funding sources.

      To be clear, we are not suggesting our findings are deterministic. There will also be, as you describe, many other facts that come into play.

      What we’ve identified is a trend: on average, amongst all the other stuff, NGOs that get more money from the government appear to devote more web-space to awareness raising and encouraging public action on development issues.

      There will always be NGOs that buck the trend, and there will always be other factors that have an influence. But there does appear to be a pattern. We’ve pointed it out and we are eager to explain it. We haven’t (as you’ll see from the blog post and the paper) drawn too many conclusions yet, other to identify the pattern and offer some possible explanations for its cause.

      While I think the strength of quantitative work is that it can sometimes identify a signal amongst the noise (what we’ve done here). I think a comparative strength of qualitative research is that it can provide nuanced explanations for trends and patterns. This, I think, would be a good way of taking this research forwards. Such research could incorporate many of the comments you have made above.

      Thanks again


  • Terence even thought the number is large (>50%) it may not impact running costs as these DFAT funds are required to be quarantined (contract wise) so the impact may not be as much as if it is general revenue. Thus the problem of drawing too much from these aggregate type analyses..

    • Hi Patrick,

      So, as best I understand your argument, you are claiming that: because the NGOs in question are less heavily dependent on DFAT for the salaries of their Australian staff, and are only actually heavily dependent on DFAT funding for their work in the field (more heavily than my initial numbers would suggest), there is no reason to anticipate them being wary of doing stuff that might displease the government?

      I’m sure there is a good reason for why the pattern we identified does not conform with theoretical expectations. But I’m equally sure the explanation you’ve just provided is not it.

      Enjoy the rest of your week.

      • Terence I think it may have something to do with values c.f Save the Children of recent times. NGOs have alway spoken out on values issues and the evidence seems to support the funding effect (in aggregate) seems to me at least, to be relatively weak. In the 1980s when they were more outspoken they even got more funds. Perverse incentives maybe

        • Thanks Patrick,

          I agree good panel data would be better than the cross section we currently have. Still the cross section is, I think, better than individual data points, and makes for a good complement to qualitative work.

          I agree with you that the nature of the issues of the day could well be important (we say as much in the blog and the paper).

          Avoiding the cases of any individual NGOs, in a broad sense I completely agree understanding issues such as values and how they might be associated with the variables we’ve studied makes a lot of sense and is the stuff of interesting future work (I think we discuss this possibility in the paper).


  • There are so many issues here Terence and Sachini. As you say, a measure of quality or degree of action would be interesting, as calling for action (especially in the aid sector) is often pretty mild and unthreatening for government.

    Unless there are particular threats in the air (as there were at some stages during the Howard government) I don’t think NGO managers are too worried about negative reactions from politicians. There may be a bit more concern however about offending DFAT staff who can make specific funding decisions, especially as a number of Australian NGOs appear to be moving into more of a contractor style arrangement with DFAT.

    My experience is that NGO managers are most likely to be concerned about not offending their individual and corporate donors and the degree and focus of advocacy is often shaped by this.

    • Thanks Garth, Great comment.

      I don’t really disagree with what you say, except that I think that activities under Take Action may be mild (there weren’t any “Chain yourselves to gateposts” buttons), nevertheless they–almost by definition–involve some degree of opposition to government policy (aid cuts, asylum seekers, climate change), so I think it fair to argue such activities aren’t wholly risk free in theory. All the more so because my guess is that NGOs’ orientation (towards activism or donations) is fairly path dependent (doesn’t change much over time), which would mean that even if the current government is fine with what they’re saying, one would think that from time to time they will pay the price for rankling politicians’ or bureaucrats’ feathers, and therefore at least be a little more wary of government money. At least, that’s one reason why I expected the relationship to be in the other direction.

      That said, the relationship we observed isn’t the expected one, and your explanations might explain why. The point about individual and corporate donors being more easily offended strikes me as very interesting.

      One area I do disagree, or at least one area where the two sides of the Tasman must differ a lot if you’re right, is to do with who NGOs would be afraid of offending (politicians or bureaucrats). In NZ it would definitely be politicians, at least at present.

      As an aside, in terms of methodology, in an email yesterday an academic made a great point that this sort of quantitative work really needs to be complemented by good qualitative work. If there are any interested aspiring PhD students out there…

      • Terence I actually wrote a book on this and tracked it over time (in particular see ch.8). There are a few factors that come into play. One is scale and that is the larger NGOs can afford to devote resources to awareness raising and they coincedently get a more of the government pie for obvious reasons. A statistical association does not necessarily suggest causation. The use of the term ‘dependency’ here is loose. Either one is dependent or one is not (like pregnancy). One cannot be a little bit dependent. Given the low level of government funding there are few if any Australian NGOs that would fold if the government funding stopped. Over the last 50 years there is less than a handful, perhaps one or two that have folded allegedly becuse of a stop in government funding but even then government funding was only part of their story. The number of accredited NGOs has dropped quite a bit over the last twenty years but none of those no longer accredited have folded. If there was more agencies dependent then gag clauses would work more. I refer in Ch 8 to when the governemnt suggested gag clauses about 10 years ago and the big agencies simply said in response they would withdraw from the funding scheme, thus ending that discussion. You see dependency cuts both ways. The other issue is the focus on internet campaigns, may not be the full picture. The church agencies in particular use church based campaigns for example, and these can be on quite a big scale.

        • Hi Patrick,

          Thanks for the comment.

          As I said to Garth, I think combining qualitative with quantitative work is ultimately the best way to undertake this sort of study, so I am interested to hear of the book (which book? the history of ACFID?). I think anyone seeking to take this work further would benefit from case histories. No disagreement there. What we’ve done with our work is something not done to date: contribute to the quantitative component of potential work in this area. Learning about specific cases tells you one very useful thing; learning about overall patterns reveals another.

          Also, you wrote: “One is scale and that is the larger NGOs can afford to devote resources to awareness raising and they coincedently get a more of the government pie for obvious reasons.”
          Indeed, if you read our original blog post you will see: “In each regression we included two control variables (whether an NGO was religious or not, and the NGO’s size in terms of revenue), alongside the main independent variable of interest…” We controlled for NGO size in our regressions.

          Then, you wrote, “A statistical association does not necessarily suggest causation.”
          We’re quite aware of that. Reverse causality is not likely to be an issue, but omitted variable bias may be. Which is why we wrote in our blog post that, “Ideally, we would have been able to add more control variables into the equation…” Having said that it is hard to think of obvious confounding factors that aren’t in the existing regression model.

          Also, you wrote, “The other issue is the focus on internet campaigns, may not be the full picture. The church agencies in particular use church based campaigns for example, and these can be on quite a big scale.”
          We, control for religious or not in our regression, which ought to address this specific trait. However, as we mention in our blog post, “The relationship might plausibly have been different had we studied other media. Because of these caveats…” So once again: we are aware of the boundaries of our investigation. We wrote about them in the blog post. Nevertheless, we think the media terrain we did cover is a very important one for NGOs.

          And you wrote, “The use of the term ‘dependency’ here is loose. Either one is dependent or one is not (like pregnancy). One cannot be a little bit dependent.”
          I disagree entirely. As an NGO changes from getting 5 to 25 to 45 to 65 to 85 per cent of its funding from the government, the costs of it losing government funding for the work it cares become incrementally more severe. There may be a point where the costs of withdrawal of government funding become particularly severe (i.e. end of NGO), but long before that there will be sufficiently major costs — programmes closed and staff laid off — to change the nature of the relationship in theory.

          Also you wrote, “I refer in Ch 8 to when the governemnt suggested gag clauses about 10 years ago and the big agencies simply said in response they would withdraw from the funding scheme, thus ending that discussion.”

          This is interesting to know about. Thank you.

          • Sorry Terence I thought you were aware of it. The reference is Kilby, P. 2015. NGOs and Political Change; a History of the Australian Council for International Development. Canberra. ANU Press. The key chapter on this is chapter 8 and p.157 has the case I refer to.

            The dependency argument has been peddled for over twenty years now with little evidence of it having much effect at all. The fall in numbers of accredited agencies points to Government funding being a nice bonus for many (mostly small) agencies but not enough for them to change practice to meet accreditation criteria. Most DFAT contracts are time bound and so agencies do lay off staff etc when they come to an end.

            Nowadays most agencies are careful not to risk a sharp drop in income hence they keep the level of government funding at lower level than it was in the 1980s, when they were more trusting and got their fingers burnt.

          • Thanks Patrick,

            I know of the book, but wasn’t clear if it was the work you were referring to or not. I look forwards to reading chapter 8.

            With respect to how much money NGOs get from DFAT.

            From 2014 data (the data used in our study):

            Of the top 50 ACFID members (recall our study looked at the largest NGOs), looking at DFAT funding/total revenue for international development work, the mean NGO received 24% of its funding from DFAT. The median 26%. More than 20% of NGOs receive over 50%.

            Such significant shares of revenue aren’t pocket change.

            I would have thought that losing such a large share of revenue (or money spent in the field) would cause pause for thought. Intriguingly, within the constraints of our study, it does not appear to have.

Leave a Comment