In the absence of trade protection, neoclassical trade theory would predict that firms in these sectors are likely to disappear and resources will shift to sectors with higher returns. Taking into account firms' heterogeneity in initial productivity, this article concludes that trade protection is not in the interests of all domestic firms. We find that highly productive firms – frontier firms – are negatively affected by AD protection, with productivity declining during protection. Although their paper focuses on trade liberalization, its results can be easily transposed to the AD trade defense context described in this paper.
AD protection in Europe is of a more temporary nature than in the US5 and in Europe, unlike the US, unlisted firms also disclose firm-level information on an annual basis. We identify firms in the European Union (EU)6 in sectors directly affected by AD policy and use their corresponding firm-level company accounts data to obtain output and input measures for the period 1993 to 2003 to estimate Total Factor Productivity (TFP) before and after AD protection. Firms that claim protection on average have lower productivity compared to firms in the non-AD control group.
We find that average firm-level wages increase after protection, which could be consistent with an increase in the skills mix. Even after the protection period, the average productivity of protected firms is still below that of firms in the matched control group belonging to other sectors of the economy. An alternative scenario in which without anti-dumping protection the low-productivity firms would have left the market and resources would shift to other more productive sectors in the economy where productivity levels are higher therefore seems a better idea.
When the Commission decides to "close" the AD case, the dumping complaint is rejected and the EU industry does not get further import exemptions. In order to analyze the relationship between AD protection and the productivity of EU producers, we identify 4,799 EU companies operating in the same sector as the dumped products. We collect company-level data for the EU import-competing sector based on the 4-digit NACE sector in which the product studied was classified.
The next year, in 1997, a fresh petition by EU producers of "Synthetic Fiber Ropes" was initiated against India and this time the EU Commission decided to grant protection from 1998 onwards. This means that EU firms in the import competitive sector were protected from 1998 onwards. Another type of overlap occurred when two different cases are mapped to the same 4-digit NACE.
Given the large number of AD cases included in the analysis, it is not our intention to conduct an in-depth analysis by sector.
Empirical Methodology and Results
To control for pre-policy trends in productivity, we also include sector 'lagged labor productivity' as an additional variable. The YEAR dummies capture for both control and AD-protected firms any temporal effect on TFP common to all firms due to e.g. Finally, the term AD_EFFECT is a dummy value equal to 1 for years after protection and zero for years before, but only for the group of firms in sectors j that received protection.
This AD_EFFECT captures the essence of the DD approach, as the coefficient estimates the differential effect that AD policy has on protected firms versus firms in the different control groups. The corresponding control group consists of sectors that never received AD protection but with a predicted probability that was at least equal to the 75th percentile of the predicted probability of protection in the group of sectors that did receive AD protection. In addition, we propose that mean values of the explanatory variables - used in the multi-nominal logit model - of the corresponding group are statistically similar to the treatment group, the so-called balancing property.
19 In the working paper version, we report NACE sectors in the "matched" control group with OLS and O-P estimates of the labor-capital production function coefficient per sector. This can be seen by comparing TFP between groups of firms in the pre-filing period, as shown in column 1. A firm that applies for protection but does not receive it is, on average, only 65% as efficient as the average firm in the corresponding control group . companies that never invested.
A company that files a declaration and later receives protection is only about 60% as efficient as the average company in the matched control group in the period before the declaration. In the five-year period following filing, the average protected company becomes slightly more productive. Note that we cannot use unit values as deflators for the matched control group, as the matched sectors involve products other than AD products. Therefore, we can only use the 4-digit PPI deflators.
In column (1), we cannot find a significant increase in average prices as a result of AD. As discussed in the introduction, there is theoretical reason to believe that the effects of protection on productivity may vary across firms. We define the initial distance-to-the-frontier for each firm i as the ratio of TFP to the productivity of frontier firm j in the first year of our sample.
We would expect AD protection to have a greater effect on the productivity of those firms whose main and only line of activity falls in the same NACE sector as the AD activity.
In Table 7, we report the results of a differentiation-in-difference analysis with firm-level fixed effects, comparing gross investment, employment, R&D25 and wages between firms in AD protection cases and firms in redundancy situations, which may most resemble the protected companies. Protected firms pay more to their workers, which could reflect rent sharing or a change in the skills mix at the firm level, with unskilled workers being replaced by better skilled ones. While we cannot verify this in our dataset, it is clear that such a change in product mix is likely to result in higher productivity.
This paper empirically measures the effect of temporary Antidumping (AD) protection on firm-level productivity of import-competing domestic firms. While we find that the productivity of the average firm is moderately improved during AD protection, productivity remains below that of firms that have never been involved in AD cases, calling into question the desirability of protection. The effect of hedging on firm-level productivity that we find is subject to strong heterogeneity.
The productivity decline of frontier firms is an additional cost of protection that emerges from this paper that adds to the loss of domestic consumer surplus and sub-optimal levels of output. This refers to a "mixed case" in which the EU Commission accepted the price undertakings offered by some of the exporters. This case consists of 3 subjects belonging to the same sector: "Bags and school bags"; "Goods for luggage and travel"; "Leather bag".
After setting the mean TFP level in the matching group to 100, we can express the means of the closing group and the confirmation group as a percentage. For example, prior to filing, cancellation cases are only about 65% as productive as the matched group, while affirmative cases are, on average, only 60% as productive compared to matched. We define distance as the initial “distance to the frontier” for each firm i as the ratio of total factor productivity (TFP) over productivity in frontier firm j in the initial year of our sample.
This frontier firm is the firm with the highest TFP in the same 4-digit NACE industry. Note: We define the initial “distance to the frontier” for each firm i as the ratio of total factor productivity (TFP) over productivity in frontier firm j in the initial year of our sample.
Change in TFP after Protection related to initial distance
Mullainathan (2004), "How much should we trust difference-in-differences estimates?", Quarterly Journal of Economics p. , vol. Slaughter (2005), "Wages and International Rent Sharing in Multinational Firms", Review of Economics and Statistics, Vol.
Ederington in P.McCalman (2008), »Endogena heterogenost podjetij in dinamika liberalizacije trgovine«, Journal of International Economics, v pripravi. Flynn (1999), "Welfare Costs of the US Antidumping and Countervailing Duty laws", Journal of International Economics, vol. Todd (1997), »Matching as an Econometric Evaluation Estimator: Evidence from Evaluating a Jobs training program«, Review of Economic Studies, 64:4, str.
Helpman (2006), “Trade, FDI and the Organization of the firm, Journal of Economic Literature, XLIV (3), pp. Tybout (2003), “Why Plant level Productivity Studies are often misleading and an alternative approach to inference”, NBER Working Paper No. 9617. Griliches (1996), 'The Inconsistency of Common Scale Estimators When Output Prices are Unobservable and Endogenous', Journal of Applied Econometrics, vol.
Liebman, (2006), "Safeguards, China and the Price of Steel", Review of World Economics/Weltwirtschaftliches Archiv, Vol. Jaumandreu (2005), "Panel data estimates of the production function and the income function: What difference does it make?", Scandinavian Journal of Economics, vol. Feenstra (ed.), The Effects of US Trade Protection and Promotion Policies, NBER vol., The University of Chicago Press .
Van Biesebroeck (2005), "Exporting Raises Productivity in Southern African Manufacturing Firms, (2005), Journal of International Economics, 67(2), december, pp.