08 June 2011


Does framing matter?

The IFS released a report today on the UK pensioners "Winter Fuel Allowance". The report found that UK pensioners spend 40% of this allowance on fuel, even though it is a completely unrestricted source of income. This compares to the 5% of expenditure on fuel from other income sources they have.

We know that spending patterns are affected by who earns income within a household (e.g. income earned by women has a larger effect on food security and nutrition at the margin). We also know that income that can only be spent on certain items has an extra-marginal effect on the consumption of that item, (e.g. food stamps, which have to be spent on food, actually do increase food expenditures, although not by the full amount of the voucher because of reductions in food expenditure from other income sources). But when income is not restricted to any expenditure category, as in Deaton and Case's work on pension income in South Africa, the finding has generally been, all other things equal, that "a Rand is a Rand".

So, to get a massive marginal increase in fuel expenditures just by framing a general income transfer as a fuel allowance is really surprising. Or is it? The adherents to Cass and Sunstein, the "nudgers", (now there's a label that needs updating) argue that when faced with an information overload, people increasingly rely on cues from friends, family, media, and the government to help them make decisions.

Can nudging work in research uptake? We all frame our arguments, written or oral, to have maximum effect (I'm trying to do it now). Perhaps the most surprising thing is that many of us spend a lot of time worrying about whether our research will be used but not much time in experimenting about how to do it and whether the experiment worked.

One example of an experiment on research uptake is an IDS/3IE collaboration around the policy impact of opinion pieces. Using a randomised trial we are investigating the stickiness of messages from a major new report under three circumstances: (a) without an accompanying opinion piece, (b) with an anonymous opinion piece, and (c) with a named opinion piece from someone well known in the field.

As this field develops and is populated with studies, it will be interesting to observe whether decisions based on cues really are more frequent in contexts of information overload, or whether it is simply part of the human condition.

No comments: