Testing the H-O-S Model

Leontief’s Test

• 1950s: Leontief conducts the first important test of H-O-S

• Using U.S. data Leontief calculated– average amount of capital and labor embodied

in U.S. exports– average amount of capital and labor embodied

in U.S. imports

Leontief’s Test

• Presumably, the U.S. was relatively K-abundant at that time

• Therefore, according to the H-O-S model, the U.S. should tend to export K-intensive products, and import L-intensive products

• That is, the capital-labor ratio for U.S. exports should be greater than the capital-labor ratio for U.S. imports

The Leontief Paradox

• Leontief found something surprising:• (K/L)exports = $13,991 per person-year

• (K/L)imports = $18,184 per person-year• This is the opposite of what the H-O-S

model predicts• This finding came to be known as the

Leontief Paradox

The Leontief Paradox

• To see this from another angle, consider the Leontief statistic– [(K/L)imp]/[(K/L)exp]– If H-O-S is correct, this statistic should be less

than one for the U.S.• However, Leontief found the statistic to be

($18,184/$13,991) or about 1.3.

Explanations for the Leontief Paradox

• Much research since Leontief’s time has focussed on trying to explain the paradox

• Do any of these explanations “rescue” the H-O-S model, or is the model just wrong?

Explanation #1: Demand Reversals

Recall: when the K-abundant country has very strong domestic demand for the K-intensive product, and the L-abundant country has very strong domestic demand for L-intensive products, there can be a demand reversal: the K-abundant country will export the L-intensive product because it has the relative cost advantage in it.

Explanation #1: Demand Reversals

• Therefore, the H-O theorem breaks down• If demand reversals are commonplace, we

might expect the U.S. to export relatively labor-intensive products

Explanation #1: Demand Reversals

• So: is there any evidence for widespread demand reversals?– No. Demand patterns are actually quite similar, at

least among industrialized countries– Furthermore, demand reversals imply that U.S.

wages should be low. This would be a hard argument to support

• So we need to look further to explain the paradox

Explanation #2: Factor Intensity Reversals

• Recall: a FIR occurs when a good is relatively K-intensive at one set of factor prices, but relatively L-intensive at another

• If FIRs occur often, the H-O theorem cannot be valid for both countries, and so we might expect the Leontief paradox

Explanation #2: Factor Intensity Reversals

• Minhas (1962) found evidence that FIRs are fairly commonplace

• Later work by Hufbauer (1966) and Ball (1966) suggests that Minhas overstated the matter

• There may be some FIRs in the real world, but not as many as Minhas suggested

• It would seem that if there is an explanation of the Leontief paradox, it lies elsewhere

Explanation #3: The U.S. Tariff Structure

• The H-O-S model assumes free trade, but in fact there are barriers (e.g., tariffs)

• The Stolper-Samuelson theorem leads us to expect that the owners of the scarce factor will be protectionist

• In the U.S., this will likely mean that it is L-intensive imports that are being kept out

• So what?

Explanation #3: The U.S. Tariff Structure

• The tariff structure could make the Leontief statistic artificially high, and perhaps lead to the paradox

• Consider an example:

Explanation #3: The U.S. Tariff Structure (An Example)

FreeTrade

U.S. Imports (K/L)imp

Cheese $10,000

Textiles $3,000

Computers $30,000

Dog Food $15,000

Average $14,500

Explanation #3: The U.S. Tariff Structure (An Example)

FreeTrade

WithTariffs

U.S. Imports (K/L)imp (K/L)imp

Cheese $10,000 $10,000

Textiles $3,000

Computers $30,000 $30,000

Dog Food $15,000 $15,000

Average $14,500 $18,333

Explanation #3: The U.S. Tariff Structure (An Example)

• Suppose that (K/L)exp = $16,000• Then the fact that tariffs exist means that the

Leontief statistic is $18,333/$16,000 = 1.14; it would have been $14,500/$16,000 = 0.9 under the assumption of free trade

• This means that Leontief’s paradox might be the result of tariffs, and isn’t evidence against the H-O-S model

Explanation #3: The U.S. Tariff Structure

• A study by Baldwin (1971) suggests that (K/L)imp for the U.S. would be about 5% lower if we allow for the tariff structure

• This would lower Leontief’s statistic from 1.3 to 1.23

• This lessens the extent of the paradox without explaining it all

Explanation #4: Adding Other Factors of Production

• Keesing (1966) suggests subdividing labor into eight skill categories

• He found that the U.S. exports a lot of skilled labor-intensive products; it is the unskilled labor-intensive products that we tend to import

Explanation #4: Adding Other Factors of Production

• Since the U.S. is rel. skilled labor-abundant, this suggests that the H-O-S model does explain trade accurately: the Leontief Paradox disappears

• Later studies have supported this finding

Explanation #4: Adding Other Factors of Production

• Leontief (1956) and Hartigan (1981) found that adding natural resources as a factor of production eliminates the paradox

• However, Baldwin (1971) found that adding natural resources does not completely eliminate the paradox

The Leontief Paradox: The Bottom Line

• Allowing for demand reversals, FIRs, the tariff structure and natural resources as a factor of production may lessen the extent of the paradox

• Allowing for different levels of skill in the labor force does seem to eliminate the paradox

• The H-O-S model appears to be serviceable

Tests of the H-O-S Model for Other Countries

• Many studies provide support for H-O-S– Stolper and Roskamp (1961): East Germany– Tatemoto and Ichimura (1959): Japan– Rosefielde (1974): USSR

• Other studies did not support H-O-S– Wahl (1961): Canada– Bharadwaj (1962): India

More Recent Tests of H-O-S

• Stern and Maskus (1981) looked at exports and imports for 128 different U.S. industries

• They estimated the following regression equation:

(X - M) = -18.54 - 0.08K + 0.06H - 2.83L

More Recent Tests of H-O-S

(X - M) = -18.54 - 0.08K + 0.06H - 2.83L • Interpretation:

– the more K an industry uses the less is exported– The more labor an industry uses the less is exported– the more human capital an industry uses the more is

exported• This is basically the same finding as Keesing’s

More Recent Tests of H-O-S

• Harkness and Kyle (1975)– added natural resources to the regression

equation– found similar results: the Leontief paradox can

be resolved by considering other factors besides just K and L

Testing H-O-S: The Bottom Line

• The H-O-S has flaws, especially in its most simplistic forms

• It is still a model that can explain real world trade patterns