T C A Srinivasa-Raghavan: Strong report, weak foundations

LINE & LENGTH

The Raghuram Rajan report confuses correlation with causality, and uses induction and deduction inter-changeably.

Raghuram Rajan, former head of research at the IMF and currently a professor of finance in Chicago, was asked by the Planning Commission to write a report on how to reform India's financial sector. His report is highly readable and there is a lot in it that is unexceptionable.

But its chapter on macroeconomic policy suffers from a serious defect in method. This is that it is almost devoid of internally consistent logic. Even its understanding of macro theory seems flawed: how do you recommend a currency appreciation when a country is running a current account deficit? Or is it that you speak only on behalf of the financial sector, the real sector be damned?

Back in the late 1960s and early 1970s, when the empiricism had not completely overwhelmed economics, it used to be underpinned by formal logic rather than econometrics. Indeed, it was not for nothing that at the Delhi School of Economics, it was Amartya Sen who taught mathematical logic to first-year MA students. In those days a knowledge of logic was to economics what the ability to run a regression is now.

But during the 1970s development funders like the World Bank began to be forced by developed country politicians to better justify their loans and aid. This led to the boom in measuring, the before-and-after sort of thing. Thus, by the end of the 1970s, the dominance of logic as the driver of economic reasoning ceased. By the late 1980s, much of economics was being driven by data sets. It still is because being rational, economists went where the money was.

One completely unintended consequence of this change was the sharp increase in the number of persons with an undergraduate background in engineering who became economists. Thanks to the low priority that is given to social sciences in engineering, they were and are largely innocent of the ambiguities and dilemmas of the social sciences, its methods of debate, its underlying non-linearity and, above all, its techniques of reasoning.

In the context of the Rajan Report, I will discuss only the last, namely, the techniques of logical reasoning. These are, essentially, two. One is deduction and the other is induction. Both have their adherents and critics. And as the Karl Popper-induced debates eventually proved, neither is wholly right or wrong. This means that the context also becomes important sometimes. Nowhere is this more true than in macroeconomics.

In deductive logic or reasoning, if I may remind readers, the conclusion depends on the premiss. If the premiss is wrong, the conclusion will be wrong. Deduction also goes from the general to the particular. For example, if your premiss is that the private sector must, regardless of the context, play a certain type of role in the economy you will also have a second premiss that all fiscal deficits are bad because they crowd out private investment. Then since all fiscal deficits are regarded as being bad, fiscal deficits for countries in specific contexts are also regarded as being bad.

This is the standard IMF line and it has often proved to be flawed because the reasoning is not based on sound logic. It was largely as a consequence of this method of reasoning that the Asian crisis became worse than it need have been.

In inductive reasoning also there are premisses. But they don't have a causal relationship with the conclusion in that the premiss does not necessarily lead to the conclusion. Thus, if you say that all swans that you have seen are white, you can't conclude that all swans are always white. Some black swans may and do exist.

Popper almost completely undermined induction as a valid method of reasoning and reaching conclusions. He asked: Does the fact that all the swans that you have seen are white mean that all swans that exist are white? That is, could it not be that there is at least one swan that you have not seen, and that it is not white?

This is the central problem with the Rajan report: it uses induction, as when it implicitly says that since sometimes central banks succeed when they have a single target "" inflation "" they must always succeed. Indeed, the report uses deduction also sometimes. A logician, I suspect, would be most upset by its sometimes-this, sometimes-that method.

Unfortunately, and thanks in a large measure to the data boys, especially those with engineering backgrounds, this has become a generalised problem in macroeconomic analysis and policy wonking. So even though they pour scorn on anecdotal evidence, the data boys have become guilty of the same methodological flaw. They confuse correlation with causality and use induction and deduction inter-changeably.

This flaw is compounded by the very logical structure of macroeconomics. Like Euclidian geometry it is basically tautological. It makes a series of self-evident statements or only those statements none except which can logically follow from the initial conditions. The Rajan report is guilty of this last as well.

This absence of a strong logical foundation has resulted in the report being reduced to what some would call pre-conceived notions regarding how things should do. One expected better.