The above reasoning, along with the discussions of theory and USDI pattern, suggests a hypothesis and determinants of USDI and HKDI as follows:
Hypothesis: USDI in China basically is oriented toward the domestic market, while HKDI has export-oriented characteristics.
b) Labor costs: China's cheap labor should encourage FDI, especially those with labor-intensive production such as HKDI.
Correspondingly, two models are specified for the determination of USDI and HKDI as follows:
t], = ratio of HKDI flows in China to FDI in all developing countries in year t; 
Second, the symmetric sets of independent variables except Dl and D3 are employed in the models for USD1 and HKDI to see how differently the U.
Data on USD1 and HKDI in China are taken and calculated from the two publications by MOFERT (various years), USCBC (1990), and SSB (various years).
2] are 80% and 69% for the USD1 model, and 98% and 97% for the HKDI model, suggesting that the explanatory power of the models is quite good.
The similar result with regard to GDPR in the HKDI model might be seen as a result of increasing propensity of HKDI's local penetration with growing openness of Chinese market.
While the coefficient of WAGEASIAR in the HKDI model is negative as expected, and significant, the coefficient of WAGEUSR in the USDI model is far from significant even at 10% level.
The structural differences between USDI and HKDI are enhanced by institutional factors.
These factors, along with political instabilities in the late 1980s, resulted in low USDI flows in absolute amount and relative to HKDI.