Writing in a review of Justin Fox’s book The Myth of the Efficient Market, Richard Thaler (2009) has drawn attention to two dimensions of the efficient markets hypothesis, what he refers to as:

• ‘No free lunch’, what economists refer to as ‘informational efficiency’;
• ‘The price is right’, what economists refer to as ‘Pareto efficiency’.

My recent research with Carine Nourry and Alain Venditti argues that while there are strong reasons for believing there are no free lunches left uneaten by bonus-hungry market participants, there are really no reasons for believing that this will lead to Pareto efficiency, except, perhaps, by chance (Farmer, Nourry, Venditti 2012).

In separate work, I look at the policy implications of this, showing that the Pareto inefficiency of financial markets provides strong grounds to support central bank intervention to dampen excessive fluctuations in the financial markets (Farmer 2012b). These are strong and polarising claims. They are not made lightly.

Not irrationality, frictions, sticky prices nor credit constraints

Some economists will be sympathetic to my arguments because they believe that financial markets experience substantial frictions. For example, it is frequently argued that agents are irrational, households are borrowing constrained or prices are sticky. Although there may be some truth to all of these claims, my argument for direct central bank intervention in the financial markets does not rest on any of these alleged market imperfections.

Other readers of this piece will wish to challenge my view based on the assertion that competitive financial markets must necessarily lead to outcomes that cannot be improved upon by government intervention of any kind. That assertion has been formalised in the first welfare theorem of economics that is taught to every first-year economics graduate student.

The first welfare theorem provides conditions under which free trade in competitive markets leads to a Pareto efficient outcome, a situation where there is no way of reallocating resources that makes one person better off without making someone else worse off. In my work with Nourry and Venditti (Farmer, Nourry, Venditti 2012), we show why the conditions that are necessary for the theorem to hold do not characterise the real world.

We make some strong but standard assumptions:

• Households are rational and plan for the infinite future;
• They have rational expectations of all future prices;
• There are complete financial markets in the sense that all living agents are free to make trades contingent on any future observable event;
• No agent is big enough to influence prices.

Crucially, in our model, there are at least two types of people who discount the future at different rates; patient and impatient agents. We show that, even when both types share common beliefs, the belief itself can independently influence what occurs. This follows an important idea originating at the University of Pennsylvania in the 1980s, what David Cass and Karl Shell (1983) call ‘sunspots’, or what Costas Azariadis (1981) refers to as a ‘self-fulfilling prophecy’¹.

The first welfare theorem, birth and death

What could possibly go wrong when agents are rational, hold rational expectations, there are no frictions, and markets are complete? Well, the first welfare theorem does not account for the fact that people die and new people are born. In our world:

• Patient and impatient agents each recognise that the financial markets are fickle and that the value of the stock market could rise or fall;
• If the markets boom then the patient savers, feeling wealthier, will lend more to the impatient borrowers;
• If the markets crash, then the savers will recall their loans, made in better times.

But in our model environment, booms and crashes occur simply as a conse