Desktop version

Home arrow Business & Finance arrow Globalisation, comparative advantage and the changing dynamics of trade

Results

Applying the approach outlined above, we find strong support for dynamic gains at the firm level. In addition, investigations for various firm-level sector groupings show that these results are not uniform across sectors. We also find that the links among imported intermediate goods, productivity, and innovation appear to be stronger in non- OECD countries: they are thus particularly important from a development point of view.

Impact on productivity and innovation

As stated above, we utilise two measures to determine the impact of imports using the firm level data: TFP and R&D spending (to proxy innovation). Results for the basic specifications are in Table 7.2.9 Using the share of imported inputs in total inputs we find a positive and significant effect on both TFP and innovation, providing robust evidence of dynamic gains at the firm level across a broad cross-section of economies.

For the level of TFP (columns 1 and 2), we find strong evidence of productivity effects from importing intermediates and capital goods: in both cases, the relevant coefficients are positive and 1% significant. These effects are quantitatively important: assuming constant returns, a firm that increases imports of its inputs by 1% increases TFP by around 0.3%; and a firm that increases it imported capital goods by 1% is around 0.2% more productive than one that increases from domestic sources only.

The smaller impact of capital goods imports on TFP could be due to several factors. One is the difference in the timing of effects. Intermediate inputs have a more immediate impact while gains from capital investment tend to be had in the longer term. Also, it is plausible that our data tend to over-sample foreign-owned firms engaged in assembly and re-exporting activities, which may not be in the best position to reap benefits from capital goods imports.

Table 7.2. Firm-level regression results:

TFP and innovation vs. imports of intermediates and capital goods

(1)

(2)

(3)

(4)

DV

TFP

TFP

R&D spending

R&D spending

Imports / total inputs

  • 0.298***
  • (0.070)
  • 0.181**
  • (0.086)

Capital goods Importer

  • 0.167***
  • (0.059)
  • 0.057
  • (0.096)

Log(employees)

  • 0.523***
  • (0.032)
  • 0.380***
  • (0.037)
  • 0.530***
  • (0.026)
  • 0.384***
  • (0.034)

Foreign

  • 0.214***
  • (0.040)
  • 0.308***
  • (0.056)
  • -0.018
  • (0.076)
  • -0.155
  • (0.154)

N

7365

4352

14800

6997

Number of groups

230

122

406

161

R2 / Pseudo-R2

0.13

0.11

0.08

0.04

Fixed effects

Country-industry

Country-industry

Country-industry

Country-industry

Estimation is by OLS in columns 1-2 and by conditional fixed effects logit in columns 3-4. Robust standard errors (in parentheses) are adjusted for clustering by country-industry. * p<0.10; ** p<0.05; *** p<0.01.

Turning to the results for capital goods, we see that there is evidence, at the firm level, of a positive and significant impact on TFP. The result is positive but not significant on R&D spending. We hypothesize that being a foreign affiliate may account for the lack of a significant relationship between capital good importers R&D spending. Theoretical work by Rodriguez-Clare (1996) shows that foreign affiliates increase a host country’s access to specialized varieties of intermediate inputs, and this access to improved knowledge raises the TFP of domestic producers as well. Empirical findings which validate this relationship can be found, for example, in Haskel et al. (2007) who report evidence for such a relationship for US manufacturing firms and Djankov and Hoekman (1999) who find that foreign investment has a positive impact on firm level TFP in the Czech Republic.

Including a variable for foreign affiliates has a positive and significant effect on the relationship with firm-level TFP, but not R&D spending.10 The apparently limited role of capital goods imports at the firm level, on innovation (as measured by R&D spending) remains.

The lack of significance of capital goods imports on firm-level innovation may be due to the type of firms involved in both R&D and capital goods importing. Firms importing capital goods (whether they be foreign affiliates or domestic firms) are usually applying adapted technology to a manufacturing process. This implies that often the R&D expenditures are made elsewhere (in the case of foreign affiliates, the home country). While there is a trend toward the increasing internationalization of R&D activities, as of 2007, more than 78% of R&D spending still took place in OECD economies, 32% of that in the United States alone (UIS, 2010). This significant relationship between imported intermediates and R&D spending may be driven by the type of R&D spending, especially, if it differs in both substance and nature to that associated with capital goods.11

To investigate the general applicability of these results, we broke the sample into two groups: OECD and non-OECD. 12 Since the Enterprise Surveys data focus more on developing and transition economies than on OECD members, our OECD sample is necessarily small.13 Indeed, there are insufficient data available to run regressions using capital equipment imports for OECD countries, and so we present split-sample results using imported intermediates data only. It is therefore important to be cautious in interpreting these results. Nonetheless, two aspects of our analysis suggest that the link between imported intermediates on the one hand, and productivity and innovation on the other, is particularly strong in non-OECD members. First, the coefficient on imported intermediates is noticeably larger in the non-OECD regression using TFP as the dependent variable (Table 7.3). In addition, only the non-OECD regression has a statistically significant coefficient on imported intermediates when we use R&D spending as the dependent variable. Both findings highlight the importance of imports of intermediate inputs regardless of the stage of development. However, the stronger results for developing countries show the major scope for leveraging imported intermediates as a source of productivity and innovation gains that can help drive the development process.

Table 7.3. Firm-level regression results for OECD vs. non-OECD countries: TFP and innovation vs. imports of intermediates and capital goods

DV

(1)

TFP - OECD

(2)

TFP - non-OECD

(3)

R&D spending - OECD

(4)

R&D spending - non-OECD

Imports / total inputs

  • 0.213*
  • (0.115)
  • 0.300***
  • (0.077)
  • 0.112
  • (0.196)
  • 0.208**
  • (0.095)

Log (employees)

  • 0.452***
  • (0.074)
  • 0.545***
  • (0.034)
  • 0.707***
  • (0.070)
  • 0.472***
  • (0.024)

Foreign

  • 0.195
  • (0.118)
  • 0.215***
  • (0.043)
  • 0.043
  • (0.185)
  • -0.034
  • (0.083)

N

1 41 1

5954

2973

11827

Number of groups

33

197

103

303

R2 / Pseudo-R2

0.10

0.15

0.17

0.06

Fixed effects

Country-industry

Country-industry

Country-industry

Country-industry

Estimation is by OLS in columns 1-2 and by conditional fixed effects logit in columns 3-4. Robust standard errors (in parentheses) are adjusted for clustering by country-industry. * p<0.10; ** p<0.05; *** p<0.01.

To provide further detail on these results, we also run regressions separately for different sectors. To preserve an adequate number of data points for each regression, we group Enterprise Surveys industries into five sectors: textiles, leather, and garments; food and beverages; heavy manufacturing (metals and machinery, chemicals and pharmaceuticals, and automobiles); electronics; and light manufacturing (wood and furniture, non-metallic and plastic materials, paper, and other manufacturing). Table 7.4 shows that imported intermediates have a particularly strong impact on productivity in the light manufacturing and food/beverage sectors. There is also a discernable but weaker impact in electronics and heavy manufacturing. Imported capital goods, by contrast, have a strong impact on productivity in two sectors only: textiles, leather, and garments; and food/beverages (Table 7.5).

Table 7.4. Firm-level regression results by sector: TFP vs. imports of intermediates

Sector

(1)

Textiles leather and garments

(2)

Food and beverages

(3)

Heavy

manufacturing

(4)

Electronics

(5)

Light

manufacturing

Imports / total inputs

  • 0.098
  • (0.079)
  • 0.566***
  • (0.124)
  • 0.270***
  • (0.069)
  • 0.241**
  • (0.063)
  • 0.622***
  • (0.148)

Log

(employees)

  • 0.426***
  • (0.032)
  • 0.669***
  • (0.036)
  • 0.289**
  • (0.087)
  • 0.486***
  • (0.041)
  • 0.622***
  • (0.078)

Foreign

  • 0.260***
  • (0.051)
  • 0.321**
  • (0.097)
  • 0.108
  • (0.063)
  • 0.281**
  • (0.074)
  • 0.168
  • (0.108)

N

2214

1917

1235

246

1753

Number of groups

49

54

33

8

86

R2

0.312

0.203

0.028

0.615

0.112

Fixed effects

Country-industry

Country-industry

Country-industry

Country-industry

Country-industry

Estimation is by OLS. Robust standard errors (in parentheses) are adjusted for clustering by country-industry. * p<0.10; ** p<0.05;

*** p<0.01.

Table 7.5. Firm-level regression results by sector: TFP vs. imports of capital goods

(1)

(2)

(3)

(4)

(5)

Sector

Textiles leather and garments

Food and beverages

Heavy

manufacturing

Electronics

Light

manufacturing

Equipment

importer

  • 0.158**
  • (0.058)
  • 0.303*
  • (0.138)
  • 0.102
  • (0.109)
  • 0.336
  • (0.152)
  • 0.063
  • (0.220)

Log

(employees)

  • 0.348***
  • (0.035)
  • 0.635***
  • (0.063)
  • 0.271**
  • (0.093)
  • 0.417***
  • (0.048)
  • 0.292***
  • (0.073)

Foreign

  • 0.243***
  • (0.055)
  • 0.460
  • (0.274)
  • 0.296*
  • (0.107)
  • 0.453***
  • (0.045)
  • 0.186
  • (0.165)

N

1696

501

1053

607

495

Number of groups

38

17

24

8

35

R2

0.095

0.26

0.177

0.533

0.057

Fixed effects

Country-Industry

Country-industry

Country-industry

Country-industry

Country-industry

Estimation is by OLS. Robust standard errors (in parentheses) are adjusted for clustering by country-industry. * p<0.10;

p<0.05; *** p<0.01.

We find that the imported intermediate share of total inputs has a positive and significant effect in all industry segments’ TFP with the exception of the textiles grouping. It is likely this has more to do with the nature of the inputs to the textile sector than the quantity of those imported inputs. We know that the textile sector imports more intermediate inputs than, say, the food and beverages sector, yet the imported inputs share shows a relatively large (second only to light manufacturing) impact on TFP of food and beverage firms and not, as stated, on textiles. Thus, it is not just the volume of imported intermediates that is determining its impact on productivity but instead is more likely a function of the type of intermediate inputs that are imported. Much of the intermediate imports for the textile sector are raw materials which may not have the level of embedded technology as the imported intermediate inputs of other sectors do. In food and beverages, by contrast, products such as fertilizers and high-yield crop varieties can have a direct effect on productivity.

The especially strong results for light manufacturing may be explained by an economy’s ability to adopt the imported technology, if we argue that the intermediate imports of the electronics sector require more skill in integrating than those in light manufacturing. We see a positive and significant coefficient for the electronics sector, but the size of the impact is smaller than for light manufacturing (1% increase in imported intermediates share leads to an increase of 0.62% in light manufacturing TFP versus 0.24% in electronics). We present evidence below that access to skilled labour influences a firm’s ability to generate TFP gains. It could be that the type of intermediate inputs imported for light manufacturing are more easily adapted and dispersed through a greater number of entities than the technology embodied in electronics.

While imported intermediates shares are not significant in the textiles grouping, equipment imports are. This is in contrast to the other four sectors examined, each of which shows much stronger results for imported intermediate share. This implies that many textiles operations import more specialized (and thus not easily adapted and dispersed for wider gains) equipment to be used with domestically sourced (usually less- skilled) labour and may also further explain the lack of a relationship with TFP.

Tables 7.6 and 7.7 repeat the sector-specific regressions using innovation as the dependent variable. Again, results differ considerably across sectors. We find that imported intermediates have a particularly strong effect in the electronics sector, and discernable impacts in the food/beverage and light manufacturing sectors. These results are not dissimilar to those for productivity, reported above. In the case of innovation, however, we do not find any significant impact of equipment imports.

Table 7.6. Firm-level regression results by sector: Innovation vs. imports of intermediates

(1)

(2)

(3)

(4)

(5)

Sector

Textiles leather and garments

Food and beverages

Heavy

manufacturing

Electronics

Light

manufacturing

Imports /

total Inputs

  • -0.214*
  • (0.106)
  • 0.496*
  • (0.219)
  • 0.472**
  • (0.145)
  • 0.727**
  • (0.231)
  • 0.146
  • (0.189)

Log

(employees)

  • 0.428***
  • (0.036)
  • 0.530***
  • (0.055)
  • 0.614***
  • (0.054)
  • 0.476***
  • (0.139)
  • 0.587***
  • (0.051)

Foreign

  • -0.314*
  • (0.132)
  • 0.237
  • (0.144)
  • -0.081
  • (0.126)
  • 0.222
  • (0.277)
  • 0.056
  • (0.197)

N

4282

2934

4030

483

3071

Pseudo-R2

0.045

0.107

0.124

0.083

0.088

Fixed effects

Country-industry

Country-industry

Country-industry

Country-industry

Country-industry

Estimation is by conditional fixed effects logit. Robust standard errors (in parentheses) are adjusted for clustering by country- industry. * p<0.10; ** p<0.05; *** p<0.01.

Table 7.7. Firm-level regression results by sector: Innovation vs. imports of capital goods

Sector

(1)

Textiles leather and garments

(2)

Food and beverages

(3)

Heavy

manufacturing

(4)

Electronics

(5)

Light

manufacturing

Equipment

importer

  • 0.060
  • (0.166)
  • 0.170
  • (0.160)
  • 0.109
  • (0.223)
  • 0.036
  • (0.303)
  • 0.026
  • (0.196)

Log

(employees)

  • 0.246***
  • (0.053)
  • 0.363**
  • (0.114)
  • 0.510***
  • (0.056)
  • 0.415***
  • (0.081)
  • 0.485***
  • (0.077)

Foreign

  • -0.201
  • (0.182)
  • 0.696*
  • (0.281)
  • -0.059
  • (0.174)
  • -0.943***
  • (0.155)
  • 0.105
  • (0.225)

N

2361

986

1796

733

1121

Pseudo-R2

0.017

0.063

0.068

0.066

0.056

Fixed effects

Country-industry

Country-industry

Country-industry

Country-industry

Country-industry

Estimation is by conditional fixed effects logit. Robust standard errors (in parentheses) are adjusted for clustering by country- industry. * p<0.10; ** p<0.05; *** p<0.01.

 
Source
< Prev   CONTENTS   Source   Next >

Related topics