wordpress-themes.org wordpress themes wordpress themes

GLOBAL FACTOR TRADE: Implications for Net Factor Trade 2

The Simple HOV Model Employing US Technology: P1 and T1

We have the same point of departure as prior studies: an assumption that all countries share a common technology matrix and an implementation that uses that of the United States. However, our study is the first to examine directly the production component of this model. As one can see in Table 4, specification P1 fails miserably, but in an interesting way. A plot of P1 for all countries appears as Figure 3. The US is excluded, since it fits perfectly by construction. A glance at the plot reveals two key facts. First, for all countries and factors, measured factor content of production is always less than predicted. Second, this gap is most severe for ROW.
We are glad to offer quick instant cash loans that will solve all your problems. Your loan payments are withdrawn automatically from your account later, so you will be repaying your debt without even worrying about it. Apply now here get-instant-loans.com and find out just how easy and convenient it is to borrow money.

This carries a simple message: if these countries used the US technology matrix to produce their actual output, they would need much less of each factor than they actually employ. The slope coefficient of measured on predicted factor trade is only 0.24. Excluding the ROW raises the slope coefficient to 0.67, still well short of the theoretical prediction of unity. The results by factor are presented in Table 5. The median prediction error is 34 percent for capital and 42 percent for labor. Thus our direct data on production suggest strongly that adjusting for productivity differences will be an important component in getting HOV to work.

Now consider trade specification T1. A plot appears as Figure 4. Factor abundance correctly predicts the sign of measured net factor trade only 32 percent of the time. This is significantly worse than relying on a coin flip! 19 The variance ratio is 0.0005, indicating that the variance of the predicted factor content of trade is about two-thousand times that of measured. This is missing trade big-time! And the slope coefficient is zero (actually -0.0022, s.e. = 0.0048).

Since the production specification P1 performs so poorly, it is perhaps no surprise that the trade specification T1 is likewise a debacle. Nonetheless, this provides an extremely important baseline for our study precisely because it reveals that our data exhibit all of the pathologies that plague prior studies. Hence we can rule out that changes in the country sample, aggregation of many countries into a composite ROW, or the selection of productive factors suffice to account for positive results that may follow.

Examination of specification P1 strongly suggested that the US technology matrix is an outlier. Is it useful to think of there being an average technology matrix B that is a good approximation to a common technology? That is the question explored in specifications P2 and T2. If we focus first on regressions based on our ten OECD countries, the slope rises sharply to 1.27, reflecting most strongly the influence of high productivity in the US.