Here is an engineer working for NASA describing his approach:

“I try to predict what happens when a spacecraft enters the atmosphere of a planet….. We attack the problem from two sides. First, we break the problem into well-defined parts and use theory and experiment to determine specific parameters under controlled conditions…. The second approach to modelling the flow field is to determine what parameters really matter to the design…. Usually with this parametric uncertainty analysis it is possible to isolate several critical parameters that require particular attention…… we fully recognise that representation of the world will never be hundred percent accurate.”

The engineer tackles the complex real world problem by positing a series of small worlds whose behaviour can be understood and building a picture of the whole, in this way identifying the factors which really matter in understanding the performance of the spacecraft and its behaviour. This is how practical knowledge is advanced.

         We have chosen this extract because Edward Prescott uses it to make the claim that  ‘the methodologies used in aerospace engineering and macroeconomics to make quantitative predictions are remarkably similar’ The reader can judge the validity of that assertion by comparing the account by the engineer, Graham Candler, with the account of Prescott’s own work which he sets beside it.

The study that began in late 1999 was motivated by the question of whether the stock market was overvalued and about to crash. At that time people did not know how to use this theory to obtain an accurate answer to this question and relied on historical relations such as price–earnings ratios to answer the question. …the tax and regulatory system had to be modelled explicitly. For example, we set the model’s tax rate on corporate distributions equal to the average marginal tax rates on distributions. This is calibration because in the model world this tax rate is the same for all individuals when in fact it is not…….corporations have large stocks of unmeasured productive assets and that these assets are an important part of the value of corporations, being stocks of knowledge resulting from investment in research and development, organization capital and brand capital. We figure out how to estimate this stock of unmeasured capital using national account data and the equilibrium conditions that the after-tax return on measured and unmeasured capital are equal.

A theory is tested through successful use. The theory correctly predicts the great variation in the value of the stock market in relation to GDP, which varied by a factor of 2.5 in the United States and by a factor of three in the United Kingdom in the 1960–2000 period.

The contrast between the humility of Candler’s account and the hubris of Prescott’s is immediately apparent. But more importantly, even a superficial reading of the two descriptions reveals that there is no equivalence whatever between the methods of the two authors, beyond the fact that both are dealing with systems that are imperfectly understood. The engineer is undertaking empirical research to discover what works, the economist is massaging data to support a priori assertions. And more importantly still, NASA is successful in executing missions of extraordinary complexity while Prescott, despite the implicit claim, does not have the slightest idea whether the stock market is or is not overvalued at any particular time.