5.3.1 Labor’s Mobility
If labor could easily move across regions in response to trade and ODI shocks, our estimates would be biased. Artu¸c et al. (2010) and McLaren and Hakobyan (2010) both found limited migrations due to trade shocks in America. In our research, we estimated the effects of ODI and net export on the size of working-age population at county level. The population statistics of each county and each year were from the Ministry of the Interior, R.O.C., and they included individuals ranging from ages 15 to 64.
Table 6 shows the results. We used lagged terms in different spec-ifications because the labor movement as an adjustment might be a long-term process. In IV estimations, both ODI and net export are treated as endogenous. All coefficients are insignificant, showing that working-age population is not related to ODI or net exports.
We used two alternative specifications to perform robustness checks on labor’s mobility. The first one was to focus on middle-aged samples, which we expected to have more difficulty in moving across counties.
We focused on the samples of citizens 40 and above.
The second was to focus on labor with no college education. We expected such worker would tend to not move across counties. More-over, in the case of Taiwan, high-educated workers tend to go to China as part of the ODI. Under these circumstances, we could not easily identify the impact of ODI on high-educated workers in Taiwan.
The first and fifth column of Table 7 report the results for the middle-age sample. The effects of ODI and net exports are consistent with the estimations in the full sample. The second and sixth columns report the results for the no-college education sample. The effects on unemployment are consistent with the full sample.
Table 6: Population Change ln(population)
(1) (2) (3) (4) (5) (6)
ln ODIt -0.011 0.189
(0.022) (0.148)
ln N et exportt -0.005 -0.151
(0.012) (0.098)
ln ODIt−1 -0.007 0.103
(0.021) (0.105)
ln N et exportt−1 -0.005 -0.086
(0.011) (0.078)
ln ODIt−2 -0.010 0.011
(0.021) (0.079)
ln N et exportt−2 -0.007 -0.035
(0.010) (0.068)
Method OLS OLS OLS IV IV IV
Observations 299 276 253 294 271 248
Adjusted R2 0.999 0.999 0.999 0.998 0.999 0.999 Standard errors in parentheses. ∗ p < 0.1, ∗∗ p < 0.05, ∗∗∗
p < 0.01. The sample period covers from 1998 to 2010 with 23 counties. Year effects and county effects are included.
5.3.2 Trade Measurement
Hong Kong, as a special administrative region of China, plays a special role in the trade between Taiwan and China. Many of the Taiwanese-Chinese imports and exports go through Hong Kong. To speculate about the real impact of the demand from China, we therefore included the trade values between Hong Kong and Taiwan. The results are in the third and eighth column of Table 7. The effects of ODI and net exports on employment and wage are similar to those estimated with statistics including only China.
5.3.3 Alternative Instrumental Variables
Autor et al. (2013) identified the growing capabilities of China’s US-bound export industry in the two decades under investigation. Being the primary sources of intermediate goods, Taiwan’s exports and ODI could be correlated with China’s exports to the USA. This theory is supported by the unlikelihood of China’s US-bound exports being correlated with the error term in our empirical specification. This makes China’s exports to US an appropriate instrument variable. We treated ODI and net exports as endogenous, estimating them with three instruments, which were the world’s stock in China, the trade
Table 7: Robustness Check on Male Samples
unemployment ln(wage)
(1) (2) (3) (4) (5) (6) (7) (8)
ODI 0.008∗∗ 0.008∗∗ 0.006∗∗ 0.006∗∗ -0.029∗∗∗ -0.024∗∗∗ -0.024∗∗∗ -0.023∗∗∗
(0.003) (0.003) (0.003) (0.003) (0.007) (0.005) (0.005) (0.005)
ODI*Skilled labor -0.011∗∗∗ -0.011∗∗∗ -0.010∗∗∗ -0.011∗∗∗ -0.004 0.015∗∗ -0.002 -0.001
(0.002) (0.003) (0.002) (0.002) (0.006) (0.007) (0.004) (0.004)
Net export -0.004∗∗ -0.003∗∗ -0.002∗∗ -0.002∗∗ 0.012∗∗∗ 0.007∗∗∗ 0.007∗∗∗ 0.008∗∗∗
(0.002) (0.001) (0.001) (0.001) (0.004) (0.003) (0.002) (0.002)
Net export*Skilled labor 0.007∗∗∗ 0.006∗∗∗ 0.006∗∗∗ 0.006∗∗∗ 0.002 -0.008∗ 0.003 0.003
(0.001) (0.002) (0.001) (0.001) (0.004) (0.004) (0.002) (0.002)
Skilled labor -0.003 0.002 -0.002 -0.003 0.286∗∗∗ 0.186∗∗∗ 0.205∗∗∗ 0.205∗∗∗
(0.003) (0.003) (0.002) (0.002) (0.007) (0.007) (0.004) (0.004)
6 years of edu. -0.008 -0.010 -0.009 -0.009 0.138∗∗∗ 0.139∗∗∗ 0.154∗∗∗ 0.154∗∗∗
(0.009) (0.009) (0.009) (0.009) (0.017) (0.017) (0.017) (0.017)
9 years of edu. -0.016∗ -0.016∗ -0.014 -0.014 0.196∗∗∗ 0.200∗∗∗ 0.243∗∗∗ 0.243∗∗∗
(0.009) (0.009) (0.009) (0.009) (0.017) (0.017) (0.017) (0.017)
12 years of edu. -0.028∗∗∗ -0.032∗∗∗ -0.029∗∗∗ -0.029∗∗∗ 0.248∗∗∗ 0.250∗∗∗ 0.296∗∗∗ 0.296∗∗∗
(0.009) (0.009) (0.009) (0.009) (0.017) (0.017) (0.017) (0.017)
>12 years of edu. -0.036∗∗∗ -0.043∗∗∗ -0.043∗∗∗ 0.422∗∗∗ 0.439∗∗∗ 0.439∗∗∗
(0.009) (0.009) (0.009) (0.018) (0.017) (0.017)
Age 0.005∗∗∗ 0.001∗∗∗ 0.001∗∗∗ 0.001∗∗∗ 0.039∗∗∗ 0.035∗∗∗ 0.037∗∗∗ 0.037∗∗∗
(0.001) (0.000) (0.000) (0.000) (0.003) (0.001) (0.001) (0.001)
Age squ. -0.000∗∗∗ -0.000∗∗∗ -0.000∗∗∗ -0.000∗∗∗ -0.000∗∗∗ -0.000∗∗∗ -0.000∗∗∗ -0.000∗∗∗
(0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000)
Spouse -0.044∗∗∗ -0.041∗∗∗ -0.037∗∗∗ -0.037∗∗∗ 0.123∗∗∗ 0.119∗∗∗ 0.122∗∗∗ 0.122∗∗∗
(0.002) (0.002) (0.001) (0.001) (0.004) (0.002) (0.002) (0.002)
Full-time job 0.307∗∗∗ 0.307∗∗∗ 0.291∗∗∗ 0.291∗∗∗
(0.006) (0.005) (0.004) (0.004)
Experience 0.020∗∗∗ 0.022∗∗∗ 0.021∗∗∗ 0.021∗∗∗
(0.000) (0.000) (0.000) (0.000)
Experience squ. -0.000∗∗∗ -0.000∗∗∗ -0.000∗∗∗ -0.000∗∗∗
(0.000) (0.000) (0.000) (0.000)
Sample age> 40 under-college H.K. 3IV age> 40 under-college H.K. 3IV
Observations 103,646 149,791 218,698 218,698 92,591 124,123 182,260 182,260
Adjusted R2 0.016 0.015 0.016 0.016 0.338 0.271 0.354 0.354
Standard errors in parentheses. ∗ p < 0.1,∗∗ p < 0.05,∗∗∗ p < 0.01. In the third and eighth column, the values of net export include Hong Kong. The fourth and eighth column report the results of estimates with three IV. The p-values of Hansen J statistics are 0.05 and 0.15 in specification (4) and (8), indicating that the overidentifying restriction is only valid in the regressiong of unemployment.
between China and South Korea, and China’s exports to the USA.
The results are in the fourth and eighth column of Table 7, which are consistent with the results from using two IV variables.
6 Conclusion
From 1998 and 2010, Taiwan’s yearly trade surplus with China grew from less than 200 million USD to 800 million USD, and yearly ODI in China grew from 2 billion USD to 14 billion. We used the Manpower Utilization Survey pooled dataset during this period to examine the labor-market effects of the tremendous changes of the values of ODI and trade.
Our research provides measures for ODI and trade at the region level with theoretical foundations. It also demonstrates how to si-multaneously estimate the effects of ODI and trade. The empirical findings showed that, for male laborers, ODI has opposite effects on unskilled and skilled laborer’s unemployment levels, with the benefits going to skilled laborers. However, it has negative wage effects on all labor. Likewise, the trade surplus had opposite effects on unskilled and skilled laborer’s unemployment levels, with the benefits going to unskilled laborers. It also raised the wage level for all laborers. Com-pared to ODI, trade surplus has less of an impact on the labor markets.
As for wage effects, it makes only half the impact that ODI does.
The findings of our study contribute to the debate that whether the economic relations with China poses threat to Taiwan’s labor.
In our calculations, the total labor-market effects of ODI and trade could explain more than 40% of the real changes of unemployment rate and wage level. The impact on unskilled workers are more significant than that on skilled workers. Such conclusions might provide some policy implications that government should be wary of relaxing the restrictions of ODI and trade with low-labor-cost countries like China.