Lower output tariffs can provide productivity gains by inducing tougher import competition, whereas cheaper imported inputs can increase productivity via learning, variation or quality effects. Theoretical models consider both the effects of reducing final product tariffs and input tariffs on productivity. We then regress factory-level productivity on final product tariffs and input tariffs at the 5-digit ISIC level.

To see whether trade liberalization has a greater effect on importing firms, we compare input tariffs with importing firms. A decrease in input tariffs of 10 percentage points leads to a productivity gain of 3 percent on average, and a productivity gain of 11 percent for importing companies. This suggests that excluding input rates could lead to an omitted variable problem, causing the value to be overestimated.

According to Corden (1971), lower input tariffs increase effective protection,8 reducing import competition and thus could result in lower productivity. We show that it is importing firms that achieve the highest productivity gains from tariff reductions. For example, a lower input rate could reduce the incentives for companies to use more efficient techniques.

However, we expect that productivity gains from lower input tariffs will dominate any potential negative competitive effect.

## Trade Policy in Indonesia

They regress industry-level tariffs on an indicator of political connectedness and find it to be insignificant. They explain their result by arguing that in developing countries it is difficult for governments to offer favors in the form of high production tariffs because they are under the close supervision of international organizations such as the International Monetary Fund. The authors show that politically connected firms in Indonesia receive benefits through the import right.

But note that only about 1% of companies fall into this category, as every company in most product groups is allowed to import. Therefore, their study seems to suggest that the endogeneity of tariffs may not be so severe in the case of Indonesia. The potential bias due to endogeneity is also reduced because our estimates all include fixed effects, so if political economy factors are time invariant, this has already been accounted for (see Goldberg and Pavcnik, 2001).

In addition to the share of unskilled labor in total employment, we use the 1991 tariffs as instruments for changes in tariffs, as in Goldberg and Pavcnik (2005) in their Columbia study. When we exclude the four product groups for which the tariffs increase, the size of the coefficient rises to -0.69 (t-stat=42.7, R-squared=0.88). This suggests that tariffs in 1991 are indeed a good predictor of changes in tariffs during the sample period.

Another important form of protection provided to industries is through non-tariff barriers (NTBs), which are generally very difficult to measure. Most of the NTBs (43 HS codes) to be removed fell within product code 37101 (iron and steel base industries). The insignificance of these coeﬃcients may be due to the imprecise measurement of the NTBs or perhaps the NTBs have not yet been removed as the Indonesian government has until 2004 to fulfill these obligations.

Unfortunately, we could not find additional information on non-tariff barriers, so the rest of the analysis focuses on the effects of tariff reform.

## Data

These input data are of particular importance to this study as they enable us to construct an input tariff for each industry. The input taﬀ is calculated using HS 9-digit tariffs from Indonesia's Department of Industry and Trade. Using an unpublished correspondence between this HS 9-digit classification and the 5-digit ISIC from the BPS, we were able to match the international and production data.18 For each 5-digit industry, we calculated the input taﬀ based on the cost share of that input .

For example, if an industry uses 90 percent steel and 10 percent rubber, we give 90 percent weight to the steel tar and only 10 percent to the rubber tar. And we use a simple average of the 9-digit HS codes to construct final commodity tariffs for each 5-digit industry. We see that input tariffs are generally lower than final product tariffs, and all have been on a downward trend over the sample period, although the largest reductions occur from 1995.

Some of the five-figure industries needed to be grouped together; for example, it was difficult to separate rice production from other grain milling products, so these two industries were brought together. But notice that we have a larger number of input rates (277) because different industries use inputs in different proportions. 19Note that this is the correlation after the rate data is merged with the company data.

## Results

The coefficient on the interaction term between input tariffs and import share is equal to -1.5. The coefficients on input tariffs and output tariffs are the same as in column 5, where input tariffs are communicated with the importing firm dummy. Note that the estimates of the coefficients on output and input rates are very similar to those in.

In columns 3 and 4 we add input rates instead of output rates, and again there is a negative and significant coefficient. These results suggest that the benefits of lowering input tariffs outweigh any potential negative impact on competition. Moreover, the input tariff applies to the total input, regardless of whether it is a domestic input or imported input.

We interact the final goods tariffs and input tariff variables with the crisis dummy in columns 3 and 4 of Table 6. We see that the key results are robust – the magnitude of the coefficients on final goods and input tariffs remain unchanged. Looking at the crisis interaction terms, we find that the interaction on output tariffs is insignificant, but there is an additional effect of input tariffs.

We see that lowering production rates also increases labor productivity and that the effect is much greater than with the TFP. Moreover, the coefficient on the interactive term between input tariffs and importing firms is large and significant, indicating that importing firms also enjoy higher labor productivity. 3 with only the output tariff as a regressor, the coefficient is negative and significant at the 5 percent level, but once we add the input tariffs this becomes insignificant.

In column 6, we add the input rates and the Herfindahl index, and again the coefficient on output rates is more than halved and becomes insignificant. A 10 percentage point drop in input tariffs is associated with a 4.7 percent increase in productivity on average for all firms, and a 13 percent increase for import firms. That is, reducing output rates increases productivity, but the largest gains are achieved through imports.

## Conclusions

Some ISIC codes are in more than one commodity code, so we used their average price to obtain a price index based on the 5-digit ISIC product code. We then used the information from the SI to calculate the shares of vehicles, buildings and equipment at the 4-digit ISIC level. These shares are used to weight each of the individual aggregate deflators to obtain a sector-level capital deflator.

National and international returns to scale in contemporary international trade theory”, American Economic Review. Trade, wages and the political economy of trade protection: evidence from Colombian trade reforms”, Journal of International Economics, forthcoming. The Narrow Belt, Dutch Disease and the Aftermath of Mrs Thatcher: Notes on Trade in the Presence of Economies of Scale', Journal of Development Economics 27, 41-55.

Trade Liberalization, Market Discipline and Productivity Growth: New Evidence from India”, Journal of Development Eco-nomics. The effects of trade reforms on scale and technical efficiency”, Journal of International Economics. Trade liberalization and dimensions of efficiency change in Mexican manufacturing industries”, Journal of International Economics.

Manufacturing Firms in Developing Countries: How Well Are They Doing, and Why?”, Journal of Economic Literature XXXVIII, 11-44. If instead error terms are corrected for clustering at the industry/year level, all significant variables remain significant with p-values <0.05, except that the output rate in columns (3) to (6) becomes insignificant. Note that the input tariff interacted with FM and import share is significant at the 1% level, regardless of the clustering group.

Notes: Robust standard errors corrected for firm-level clustering in parentheses; If the error terms were instead corrected for clustering at the industry/year level, all significant variables remain significant with p-values <0.05. If we instead correct the error terms for clustering at the industry/year level, all variables with at least 5% significance remain so, except that the output tariff becomes insignificant and the input tariff interacted with the dummy crisis becomes insignificant.* significant at 10%; ** significant at 5%; *** significant at 1%; (1) crisis dummy = 1 in 1997 and 1998; (2) column 3 is FM importing firms, an import dummy, and column 4 is the interaction with import share. Annual fixed effects yes yes yes yes yes Industry fixed effects yes yes yes yes yes.

If the error terms were corrected for clustering at the industry/year level, all significance levels remain unchanged, except in column 5, the manufacturing tariff becomes insignificant and the foreign share becomes significant at the 10% level. If the error terms were corrected for clustering at the industry/year level, all the significance variables remain significant except that the output tariff in the first difference model in columns 3 and 4 becomes insignificant, the input tariff in levels and the first difference become insignificant but remains significant in the differenced model of period 5, and the Herfindahl index is insignificant in all specifications.