How the "Masters of the Universe" think about Exchange Rates!

In a recent excellent, if perhaps deliciously short, report issued by the most profitable bank this year entitled "The Foreign Exchange Market" -- one of the sub-reports attempts to forecast the exchange rates for the coming year. What struck me about it is -- other than the predictions they make (I have my issues with that... see below) -- is the elegance of the underlying method. The way do is as follows:

1. Changes in Terms of Trade (= price of exports divided by price of imports) is a function of changing commodity prices (energy, industrial metals, agriculture, live stock).
2. Extract sensitivity estimates (the coefficients in a regression) to predict terms of trade.
3. Changes in Real Exchange Rates ( = price of one unit foreign currency in domestic currency * ratio of foreign and domestic price levels) is a function of two key parameters.
   
• Terms of Trade
• Relative productivity levels -- measured by, say, per-capita output per hour etc.,
  
4. Perform regressions on #3, using #2 to arrive at new estimates for real-exchange rates.
5. Convert real exchange rates into nominal exchange rates.

Amongst key predictions are USD-CAD = 1.10; USD-INR = 50.1. i.e., their model supposedly predicts that the the Canadian dollar is expected to depreciate from the present levels, and so is the USD expected to appreciate against the Indian Rupee. Since, they do not explicitly mention all the concerned control variables in the exchange rate attribution -- it is difficult to really validate their claims, even intuitively.

My own guess is that there are three key parameters that affect the short term exchange rate fluctuations:

1. Global capital flows -- that chase the second-order effects anticipated changes in terms of trade.
2. Changes in US deficits (budgetary and trade) -- this is particularly accentuated by the coming US electoral-cycle.
3. Idiosyncratic events -- particularly emerging market macroeconomic instabilities.
   

So, I suspect their analysis are largely driven by "true" long term economic factors, while the intermediate fluctuations are more complicated beasts -- and therein lies, as Shakespeare writes, the rub.

Lessons from Leamer (Part 1)

Professor Edward Leamer has written a very interesting paper; and was presented at the Jackson Hole conference. I have been going through it in detail -- and I must admit it is a remarkable piece of analytical detective work. Following two posts are dedicated to his work. Although the original is nearly 70+ pages -- it is written in a very reader friendly style. So, worth the effort. For the rest, here is a quick summary. Caveat Lector!

1. Conclusions can be drawn of effects of policy X, if there are control and treatment groups. In macroeconomics, we only have non-experimental data. So, we rely on “story-telling”.
2. Monetary policy affects economy via housing-related decisions. Tempering business cycles means figuring out how housing-related choices are affected.
3. Inflation is a “daily” phenomenon, while housing is reinforced by a “wave”.
4. Monetary policy at different stages of the housing cycle – can exacerbate or diminish the extent of hurt that happens when a boom declines.
5. Housing booms are very susceptible to interest rate changes at the end of the boom, rather than in the early and middle stages.
6. Historically, the US economy has been growing at 3% over the past 30 years – despite all kinds of real, monetary and technological shocks. Monetary and fiscal policy must restrict itself to smoothening out the cycles (the amplitude and the frequency).
7. Contributions to long run growth are led by, in order of importance from a 2005 perspective, Consumer services, Non Durable Consumer Spending, Durable Consumer Spending, Equipment, State and Local Expenditure, Defense expenditure, Residences etc., i.e., Residences do not contribute substantively in long term growth.
8. For business-cycle fluctuations, Residences are very important. Typically, Residences contribute to the weakness in GDP growth before the “recession” starts and contribute above normal before the “recession” ends. This is in contrast to equipment and software sales. i.e., Residential spending is a pro-cyclical predictor of GDP growth. Equipment sales just mirrors GDP growth rate.
9. To address decline in GDP growth, one must address the “consumer” side of the equation and not the business side. So, Monetary policy must explicitly take into account Residential and Durables levels and volatilities.
10. There have been 10 U.S. recessions. Only two were driven by “production” side (i.e., by lack of demand on the business side) – post-Korean war and 2000/01 internet bubble collapse. Rest of the recessions have been caused by lack of consumer demand.
11. Housing prices are sticky downwards – i.e., prices don’t fall all that much. So, the natural way the market clears is by massive reduction in demand. This reduction is demand affects GDP/employment.

India -- Some Graphs