Monday, February 08, 2010

The Epistemology Of Economics

A reader (contributor) recently asked me what principles should be used in identifying just what economic indicators one should focus on when analyzing economic events. That was one of the hardest but also one of the best questions I have received.

The reason why it is hard is that there actually is no good answers in the sense of indicators which are always relevant. Some are relevant a lot more frequently than others, but there is no universal indicator which you should always focus on.

The choice of indicators is essentially contextual. By contextual, I mean the indicators most relevant in the context which is discussed. But how do you determine what is most relevant? That depends on what you want to find out, of course. But generally, one should focus on 1)The key causal factor(s) 2) The key present (or immediate) symptom(s) 3) The key likely future (or later) effect(s).

That could potentially mean almost an unlimited number of numbers, but one should "economize" according to an epistemological principle which the reader in question, and probably many others are familiar with:

"concepts are not to be multiplied beyond necessity—the corollary of which is: nor are they to be integrated in disregard of necessity."

For example, with regard to trade, when discussing factors affecting exchange rates, it is rational to simply rationalize away exports and imports and simply use the trade balance. But when discussing effects on real output, a drop in both imports and output is likely to be associated with (and partially cause) a drop in real output.

I can't list in a single post all factors of all three mentioned kinds in all situations, but one example should do: The recent financial crisis in America. What was the causal factors? As a basis, one could use my November 2004 article (suggested by the questioner) predicting the crisis. While my particular choice of indicators wasn't necessarily the best possible, I can say even in hindsight that they were good (in hindsight it probably wasn't the best possible, but it was still good). As an example of how indicators should be chosen, it will do.

The causal factor behind the housing bubble was the low interest policy pursued by the Fed.

This created the housing bubble whose immediate consequences included higher house prices, higher level of construction spending, higher levels of household debt, a bigger current account deficit, and a higher private sector financial savings deficit.

The later effects included a retreat from the excessively high levels of house prices, household debt, current account deficits and private sector financial deficits and an unsustainably low household savings rate. And since there are rigidities in the real world economy (both in the form of price rigidities and the difficulty in many cases of using workers and capital equipment for new tasks) this will ultimately mean that employment and the real value or output will decline.

Going through the different indicators,it can be asked why were they chosen:
Starting with initial interest rates it is an indicator directly controlled by the central bank.

What about the indicators of immediate consequences: Since houses are mostly a capital good, and since the present value of capital goods (which by definition generates its value at a later point in time) increases when interest rates (the price of time) goes down, it follows that house prices should increase. And since house prices are relevant for both the profitability of house construction as well as the willingness of home owners to take on additional loans this is clearly an indicator relevant for economic analysis.

Another relevant data was the dis-aggregation of household asset and data values, pointing out that while many households had big assets and almost no debts, others had limited assets and debts almost as large as their assets, making them highly vulnerable for asset price declines.

I also mentioned as a factor driving the boom, the shift from a budget surplus under Clinton to a large deficit under Bush. While this wasn't related to the housing bubble, it wasn't a sound source of demand. Another factor I mentioned was the purchases by Asian central banks of U.S. Treasuries. That wasn't however meant to represent a causal explanation of the crisis, instead it explained why the bubble persisted even though money supply growth in 2004 had moderated compared to the high levels of 2001-02. Since the absence of these purchases would have likely been associated with higher money supply growth, it wasn't really a causal explanation though.

In short: the key is to focus on the indicators that are most relevant for causing a certain event, the indicators that best reveal the existence of a certain scenatio taking place right now, and the most likely relevant effects. And when chosing indicators in these 3 stages use the "Rand's razor" to achieve understanding of what is going on.