Debt, Growth, and the Illusions of Social Scientism

Everyone interprets the world by applying a theory

by Tomas Salamanca

For all the politicians and economists who have been doggedly nonchalant about escalating levels of public debt, this was a good week. Making their week was the revelation that the statistical calculations in an influential paper were off. It is not often that a math error causes such a mix of glee and consternation as we have seen over the past several days. But the computations in question, advanced by Kenneth Rogoff and Carmen M. Reinhart, both professors at Harvard, are widely thought to have swayed governments in the Western world towards austerity policies. Thanks to a paper out of the University of Massachusetts, the empirical grounds for those policies have allegedly been swept away. Actually, what the Rogoff-Reinhart affair demonstrates is that empirical evidence does not, and indeed cannot, decide economic controversies.

That it can is the defining conceit of the neo-classic orthodoxy in contemporary economics. In this, the economics profession has long been following the dominant trend in the social sciences where a philosophic commitment to positivism became firmly ensconced in the post-World War II era. Positivism is the view that the truth of statements concerning the world we experience through our five senses can only be established by passing the test of observation. It is a philosophy influenced by the enormous success of the natural sciences in comprehending the physical and non-human universe.

Its proponents in the social sciences believe that this accomplishment holds the key to eventually securing important truths about human affairs. If the testing of hypotheses through experiments and the analysis of data is what finally advanced our understanding of the natural order after millennia of idle speculation and disagreements, then that method of inquiry is our surest bet to replicate this intellectual achievement in the human order. Because so much of their subject-matter is reducible to numbers, economists have been more disposed than the general run of social scientists to adopt sophisticated mathematical techniques and models in generating empirical tests.

Though the quantitative approach taken by Rogoff and Reinhart was relatively straight-forward – what they basically did was sort the historical data into four different groups by level of public debt and then calculate the average level of real GDP growth for each group – they did go the positivist route of testing a hypothesis. Their hypothesis was that, beyond a certain level, public debt would exhibit a significant relationship with economic growth. Using figures from 1946 to 2009, Rogoff and Reinhart originally found that, on average, real growth falls 0.1% per year when the public debt of an advanced economy exceeds 90% of GDP. That compares to an annual growth rate of 2.8% at debt levels between 30-90% of GDP and 4.1% when debt was under 30% of GDP.

What the professors at the University of Massachusetts found after re-analyzing the Rogoff-Reinhart data set was that growth for advanced countries with debt greater than 90% of GDP was 2.2%. They point out that this was not significantly different from historical instances when a country’s debt was below 90%. They chalk up the original error to the omission of key data, the use of unconventional statistical methods, and, most embarrassingly for Rogoff and Reinhart, mistakes in coding formulas within Excel.

Obviously, whether or not positivism offers the right epistemology for the study of human action is not something that stands or falls on the susceptibility of data analysis to software coding errors. But it does hinge on that data being simply a record of past events. To decipher a previous connection between A and B does not logically entail that A is causally related to B. It is merely to say that A was historically associated with B. This history is equally consistent with the notion that A is the cause of B, or B is the cause of A, or that the occurrence of the two is nothing more than a coincidence. In order for the relation to be established as causal, its necessity must be demonstrated – that is, one must show that A and B must go together. A historical analysis cannot do this without presupposing that the future will be like the past. But there is no way to prove this because the future is, by definition, something that has yet to occur. What is to come is always going to be unavailable for empirical verification at whatever instance one is trying to support a hypothesis. Neither the Rogoff-Reinhart paper, nor that out of the University of Massachusetts, overcomes this fundamental shortcoming of the empirical-positivist approach. Instead of doing economic theory, they are both actually engaged in the discipline of economic history, albeit expressed in numbers as opposed to words.

Another problem in trying to bring the methods of natural sciences into economics comes from the causal density present in social life. Exceedingly rare is the human event that can be reduced to a single cause or even a double or triple one. Far more typical is the situation in which there are a multitude of causes, each determining the effect with varying degrees of influence, with some of these subject to feedback loops. No statistical method can control for all this. It is interesting that both the Rogoff-Reinhart and University of Massachusetts studies do little to control for all the variables that go into generating economic growth except to distinguish between advanced and emerging market economies – as if all advanced economies can be lumped together in terms of tax rates, levels of government spending, protection of private property rights, soundness of the monetary system, and the myriad of other factors that determine a nation’s wealth.

Only a theoretical approach employing the rules of deductive logic can avoid these shortcomings and offer an authentically scientific account of the causal forces that operate in human affairs. As the Austrian tradition well teaches, this is done by starting with the self-evident premise that human beings act to improve their condition as they subjectively perceive it and then progressively deducing the necessary implications of this in economic life through a chain of reasoning. The resulting theories are then illustrated in the empirical record. This does not mean, as is commonly alleged against the Austrians, that theory is forced onto the facts. Theory must adequately explain the facts. If there are features of the observable story that cannot be accounted for by the theory, then we must look for another one that provides a more encompassing explanation. Even in this case, though, the facts do not disprove the theory, for the validity of the latter is based on its internal logic. Rather, the theory just doesn’t suit the facts.

Arguing in favor of this theory first mode of inquiry is that it matches the way we all intellectually grapple with reality – including positivist economists. Take Paul Krugman. In his commentary on the Rogoff-Reinhart affair, he tells us that he never accepted the implications of their paper that a ratio of public debt to GDP higher than 90% represents a drag on economic growth. He is also quite sure that now that the Rogoff-Reinhart numbers have been shown wrong that the critics of high public debt will nevertheless persist and look to other studies to defend their opinion. Krugman is right about this, but he lacks the self-awareness to apply this insight to himself: virtually no one changes their world-view because of the results of a social science study. Instead, everyone interprets the world by applying a theory. Krugman’s theory just happens to be Keynesian, whereas those of us who happen to think that current levels of public debt sorely need to be reduced are operating with a different theory.

Best to recognize this by making sure we have the right theories in our head, rather than entertaining the vain hope that one day, via some controlled experiment or the statistical parsing of voluminous data, the world can by itself tell us what to think about the human condition.