The equity-efficiency trade-off and simplifying assumptions

Given my admission that I am now going to talk more about inequality, it is important for me to show a bit more analytical respect to the concept of the ‘equity-efficiency trade-off’.  This is a term that is often used in economics, and that we often use here, but which on the blog I have only explicitly dug into once before – back in 2008.

The reason I often prefer not digging myself into the equity-efficiency trade-off concept too much is that I fear I won’t dig myself out, and if I do I doubt much would come from it.  It is an overarching concept that exists in economics, one that we have to be sure we consider whenever we ask a specific question.  However, without reference to a question there isn’t terribly much to say.

When it comes to the equity-efficiency trade-off associated with policy and social organisation, it is clear that we cannot clearly separate individual concepts associated with fairness – ideas of inequality and poverty will be inextricably linked, one of the key reasons why I dislike to push by the Spirit Level to solely place focus on inequality.

Building the trade-off

When I read literature by economists, I like to keep an eye out for the question they are using for motivation, the question they are directly answering (which is much more narrow in scope, and its relation to reality), and the simplifying assumptions they are making to link the two.  In this way, the question that is answered is indicating a “tendency” that may exist in reality – and the simplifying assumptions exist to fill a number of roles in this case.  The more we can “loosen” these assumptions, or show that they hold empirically, the more confidence we can have that this tendency may hold.

Of course, there is the additional “layer” of heading from the observable variables we have to the unobservable level of welfare we care about (as mentioned last time – and which Dalton puts nicely).  But this holds under the same type of logic!

However, this should make clear that the link between literature and reality is sequential in nature – and we need to peel back forms of simplifying assumptions in order to reach our final policy related goal of the equity-efficiency trade-off that prevails in society.  In our previous discussion we noted a series of simplifying assumptions that are used to help make specific points – and how as a result, the concept of inequality used by economists when putting together these concepts and measures, and the concept of inequality that members of society get grumpy about, are different things!  There is some relationship – and having a clear view on this relationship is required, and should be explicit and transparent before we use it to make policy!

The equity-efficiency trade-off we are interested in needs to be defined clearly.  This was, in many ways, what I was trying to get across when discussing how we use simplifying assumptions – and how communicating ideas with the public differs from communicating with economists – in this.

The example of inequality, and the efficiency-equity trade-off:

As we noted here, and in an earlier post, Dalton 1920 stated that when thinking about inequality and welfare we have to conceptualise a functional relationship between the two.  The corresponding relationship that he assumed, and that was assumed by Aigner and Heins (1968) and Atkinson (1970) (REPEC) was that we had an “additively separable and symmetric” welfare function.  In such a framework the way peoples welfare influences others, and the relative interpersonal comparisons involved, are set in a way that is unrealistic – it is a simplifying assumption!

This is fine, this set of literature then burst into life – suddenly there was a research programme filled with assumptions that needed loosening, we needed to know what parts of the core results would be sensitive to certain assumptions, and what lessons could be generalised to a broader world (this is the Credible World view of what economists do).

So far so good.  The ideas of differing needs and values made their way into the literature in the way we’ve viewed “characteristics”, and a bunch of work on decomposition and (arguably) equivalization has helped out.  But when it comes to policy conclusions, the logic of these simplier models is often leant upon.

The response of economists had been to discuss a trade-off.  We have presumed that equality is good (although as we pointed out earlier this is actually a bit of an uncomfortable assumption), but that there is a trade-off – an “equality-efficiency trade-off”.  Many papers directly say this, and some term equality = equity.  The efficiency loss of an “equality improving policy” is then viewed largely through any impact on labour supply, or the willingness to accumulate capital.  In that regard we start to feel confident we can envision an equity-efficiency trade-off – in fact the Atkinson Index (again from Atkinson 1970) can be seen as giving us a way of telling the “costs” of inequality, and the lower “value of income we would accept” in order to have a more equal distribution of income.

This rhetoric pervades the literature, and although the economists involved have a nuanced view of the role of policy, you can forgive the fact that this “sounds” like it is the end of the story.  However, it is not.

“Equity” is more than a single variable

Giving a two-dimensional view of welfare provides the idea that the trade-off is solely between equality and efficiency.  This is a lot better than ignoring the link, and I think research into the size of the trade-off or whether it even exists makes sense.

However, when the Spirit Level authors thought they had discovered there was no trade-off there was an open presumption that this was case closed – time to throw around equality policies to see what happens.  But even if they did have a compelling case for their conclusions (ones that I haven’t seen) good policy would not take some welfare function that takes aggregate equality and aggregate output as factors and then “maximise” it.  This isn’t just about the difficulties of appropriate estimation of the parameters given these assumptions (discussion here and here) – it is a broader point about the heterogeneity of individuals in terms of their ability to produce welfare/satisfaction/well-being (both for the same level of income and leisure, and the relative price of doing so).  To quote from Sen again:

As a result of that assumption, we are made to overlook the substantive inequalities in, say, well-being and freedom that may directly result from an equal distribution of incomes (given our variable needs and disparate personal and social circumstances).

These questions are complicated, and require deep and wide-ranging analysis, as Treasury has been trying to articulate with its discussion of a living standards framework.  The Stanford Encyclopedia of Philosophy also has a neat discussion on the fairness principles that go into differing discussions of distributions of income – and how often individual concepts we see as self-evidently just will conflict with each other!

Careful analysis is important – or else we will end up trying to justify policy that is analytically filled with simplifying assumptions – policy that will as a result inappropriately articulate the equity-efficiency trade-off at stake.  This is a fancy way of saying that such policies will end up creating harm.

As a result, discussing this trade-off involves empirical analysis of distributions along a number of variables and characteristics, combined with a clear way of thinking about counter-factual worlds AND a way of getting individuals in society to reveal their preferences over these things.  “Equity” does come from asking how much income society would be willing to sacrifice for non-income concepts – but this is not just a matter of equality.

Note:  The discussion of income inequality here has been premised on actually having measures of lifetime income – in reality such measures are often not there, and the actual data we are working with is imperfect.  These were issues we touched on more in this prior post.  This makes thinking about the limits to knowledge when we think of policy as well 😉