OMITTED VARIABLES, CONFIDENCE INTERVALS, AND THE PRODUCTIVITY OF EXCHANGE RATES

Abstract.  A new analytical technique, named 'Bi-Directional Reiterative Truncated Projected Least Squares' (BD-RTPLS), produces reduced form estimations and less than half the error of OLS when there are omitted variables that interact with included variables. This paper develops confidence intervals for BD-RTPLS and uses BD-RTPLS to estimate the relationship between the exchange rate (e) and gross domestic product (GDP) using annual data from 1984 to 2000 for 23 developing Asian and Pacific countries. BD-RTPLS produces estimates for the exchange rate multiplier ( d GDP/ d e) for these countries and shows how omitted variables affected these multipliers across countries and over time.     

How China＇s Holdings of Foreign Reserves Affect the Value of the US Dollar in Europe and Asia

By January 2009, China held almost US$2tn in foreign reserves. The present paper estimates the marginal effect of China changing its holdings offoreign reserves on the value of the US dollar in Europe and dsia. Because using traditional techniques to find this estimate would be inappropriate due to severe problems resulting from omitted variables, the present paper uses a new approach, bidirectional-reiterative trucated projected least squares, that has been proven to minimize problems associated with omitted variables. It is found that if China would sell 1 percent of its foreign reserves, then the value of the US dollar would fall by 0.44 percent. With such a large effect, China has an incentive to either not sell arty of its US dollar reserves or sell all of its US dollar reserves.     

China’s fiscal stimulus package for the current international crisis: What does 1996–2006 tell us?

This paper uses Bi-Directional Reiterative Truncated Projected Least Squares (BD-RTPLS) to estimate annual ∂GPP/∂G and %∂GPP/%∂G multipliers for China’s 31 provinces between 1996 and 2006 (GPP = Gross Provincial Product and G = government spending in the province). In order to account for the influence of omitted variables, BD-RTPLS produces a separate ∂GPP/∂G and %∂GPP/%∂G estimate for every observation in the data set. I find that ∂GPP/∂G and %∂GPP/%∂G were highest in 1996 while China was drastically cutting government spending to curb inflation, were lowest in 1997 during the Asian Financial Crisis, slowly rose between 1998 and 2004, and then declined some in 2005 and 2006 when the Chinese yuan was slowly appreciating.