Master thesis II, 15 hp Master’s Programme in Economics
Spring term 2021
EXPORT PARTICIPATION AND PRODUCTIVITY
Evidence from the Ghana manufacturing industry
Sarah Dawu
Abstract
Theoretical models predict that firms self-select into export market based on their productivity because only the most productive firms can overcome the entry cost in the export market. The objective of this study is to test this for Ghana manufacturing firms from 1991-2002 by studying whether more productive firms are more likely than less productive firms to start to export.
The findings indicate that firms that initially are more productive are more likely to enter export markets, which is consistent with the self-selection hypothesis. Firm characteristics also affect the decision to export; firms that are large, young and firms that have a higher share or workers that completed secondary education are all more likely to export. Moreover, firms that engage in exporting activities have shown superior characteristics in terms of number of employed persons, wages paid to employees and capital intensity, over firms that do not export.
Keywords: Ghana manufacturing industry, Heterogeneous firms, Productivity, Self-
selection, Export, Panel data.
Table of Contents
1. Introduction ... 1
2. Literature review ... 3
3. Stylised facts ... 7
3.1 Data source ... 7
3.2 Export trend and policies ... 8
3.3 Share of firms ... 9
3.4 How do exporters differ? ... 14
4. Methodology ... 16
4.1 The probability of becoming an exporter ... 16
5. Results and Analysis ... 19
6. Conclusion ... 22
References ... 24
Appendix ... 27
1
1. Introduction
Empirical research emphasises that firm heterogeneity in product portfolios generates international trade and induce aggregate productivity growth. Firms play important role in mediating countries’ imports and exports. However, in issues concerning trade, economist usually place emphasis on comparative advantage, increasing returns to scale, and consumer love for variety whiles they give little attention to the firms that drive trade flows (Benard and Jensen 1995).
Trading firms differ substantially from firms that serve the domestic market. Across a wide range of countries and industries, firms that export have shown to be larger, more productive, more skill – and capital – intensive and to pay higher wages than non- exporting firms (Bernard and Jensen 1995). These differences exist before exporting begins; evidence indicate that firms entering export markets grow substantially faster in employment and output. The fact that exporters have a productivity advantage before they start to export suggests self–selection. Exporters are more productive not only because exporting can enhance productivity, but also because merely the most productive firms are able to overcome the costs of entering export markets (Bernard and Jensen 1995).
In this study, I use Ghana manufacturing firms’ data from 1991-2002 to investigate whether more productive firms in the manufacturing industry are more likely than less productive firms to start to export. The focus on firms makes this study interesting, as other studies on exports within the West African region focus on countries exports rather than firms’ exports. The paper contributes to the previous literature by, according to the best of my knowledge, being the first to analyse this for Ghana. To give a clear understanding of the data, I first explore the differences in the characteristics between exporters and non-exporters even though the main objective of this study relates to estimating how pre-exporting characteristics predicts which firms that start to export.
I find that exporting firms are larger in size (in terms of number of employees) and more
capital intensive relative to their counterpart non-exporting firms. In addition, exporting
firms, on average, pay wages that are about 42% higher than those paid by non-exporting
firms. With respect to the objective of this study the clear result is that firms that enter
export markets show higher initial productivity compared with non-exporters; in other
words, more productive firms in the manufacturing industry are more likely to start to
2
export than less productive firms. This result is consistent with the self-selection hypothesis.
According to research studies there are higher returns available in the world market, which may induce firms to increase their productivity by investing in physical capital before they attempt to enter export, a process known as conscious self-selection
1(Lopez 2004). Knowing if the firms self-select into the export markets based on productivity is important for policy making decision. Policies that reward exporting after entering export (ex-post) may increase productivity and innovation for current non-exporters and successfully increase economic growth.
2According to Helpman et al., (2004), every industry is populated by heterogeneous firms, which differ in productivity levels. As a result, firms sort according to productivity into organisational forms. The least productive firms leave the industry because they cannot generate positive operating profits no matter how they organise. Other low productivity firms choose to serve only the domestic markets. He noted that the mode of operation in foreign markets differs, however. The most productive firms in the group choose to serve foreign market through foreign subsidiaries by engaging in foreign direct investment (FDI)
3while the less productive choose to export.
In theory of heterogeneous firms, two hypotheses have been formulated to explain the productivity premium of exporters. The first hypothesis points to self-selection of more productive firms into export markets. The reason for this is that there exist additional costs of selling goods in foreign countries. The range of extra costs include transportation costs, distribution or marketing costs, personnel with skill to manage foreign networks, or production costs in modifying current domestic products for foreign consumption.
These costs provide an entry barrier that less productive firms cannot overcome.
1 Lopez (2004) develops a model in which profit-maximising forward-looking firms invest in new technology with the intent of becoming exporters, and the adoption of the technology requires mastery and learning that only initially more productive firms can accomplish. He therefore argues that there is indeed self-selection, which involves a conscious decision to increase productivity.
2 Policymakers must act cautiously in giving rewards to firms after becoming exporters because according to empirical evidence, not all firms remain exporters. Some firms enter exporting then later become quitters (Wagner 2007).
3 There can be horizontal or vertical FDI. Horizontal refers to investment in production facilities abroad that designed to serve consumers of host nation. Vertical on the other hand is when a parent firm invests in a production facility in another country to produce inputs that will be shipped back to the parent firm for further processing. Or when a parent firm produces only inputs in the home country, and it invests in an assembly facility in another country to which it ships the inputs.
3
Moreover, the behaviour of firms might be forward-looking in the sense that the desire to export tomorrow leads a firm to improve performance today to be competitive on the foreign market (Wagner 2007).
The second hypothesis is learning-by-exporting. This explains that knowledge that flows from international buyers and competitors help to improve the post-entry performance of export starters. Furthermore, firms participating in international markets are exposed to more intense competition and must improve performance faster than firms who sell their products domestically only (Wagner 2007). Here, firms that engage in international trade become more productive after they begin to export.
The next section covers literature of the study. Section 3 entails stylised facts about the sample and policies implemented over the sample period. In section 4, I present the model of the decision of becoming an exporter and further discusses alternative methodologies for estimating the binary panel data model with firm effects. Finally, sections 5 and 6 provides results and conclusion of the study, respectively.
2. Literature review
Export represents a flow of income into an economy, increasing wealth and standards of living. Moreover, exporting firms are frequently involved in higher value-added activities (Robson and Freel 2008) and to a greater extent represent the relative competitiveness of national economies (Robson and Freel 2008). Aitken, Hanson and Harrisson (1997) report evident that plant size and wages are positively related to export and these characteristics are also associated with productive plants.
Detailed research on the relationship between export and productivity show that firms that export have higher productivity and often higher productivity growth rate and this finding tends to hold after controlling for observed plant characteristics, industry, and size. Also, future exporters tend to be more productive than future non-exporters years before they enter the export market and have higher ex-ante growth rate of productivity.
The evidence is the more productive firms self-select into export markets while, exporting
does not necessarily improve productivity (see e.g., Greenaway and Yu 2004).
4
Wagner (2007) explained that if better firms self-select into export starting and if therefore, today’s export starters are better than today’s non-exporters, it would be expected that the export starters should on average perform better in the future even if they do not start to export today. He emphasised, since they start to export today, we do not observe what would have happened if these better firms had not started exporting today. For this reason, we are unsure to say that the better performance of export starter compared to non-exporters is to some extent caused by exporting or not. What this implies is that we cannot not say for a fact that a firm has high productivity only after it entered export (learning-by-exporting hypothesis), because even before the firm starts exporting it must already be productive.
Most studies have found strong evidence to support the self-selection hypothesis while weak or no evidence to support learning-by-exporting hypothesis. Another finding is that stopping to export tends to be accompanied by a decrease in productivity in most cases.
For instance, Baldwin and Gu (2003) found that Canadian firms that exited the export market between 1974 and 1996 were 13% less productive than firms that continued in the export market.
Discussions of the role of exports in promoting the growth of an economy and
productivity have been ongoing for many years. For instance, economic leaders are
convinced that exports are good, and exporters are good firms, thus helping domestic
firms export is good policy, but exporting is a rare activity for firms. Until recently,
empirical research in the field of economics used data in the industry level to investigate
whether exports promote productivity growth or vice versa. Accordingly, Bernard and
Jensen (1999) used 1984-1992 data from the US to analyse the sources of the substantial
performance advantages at exporting plants and firms. They found that at any point in
time exporters produce more than twice as much output and are 12%-19% more
productive. In addition, exporters pay higher wages to all types of workers. When they
looked at the characteristics of plants before they export, it was clear that good plants
become exporters. Also, future exporters have performance characteristics years before
they ship any goods abroad. The plants are already characterising with faster growth for
export starters than their non-exporting counterparts. They concluded that success and
new products lead to exporting and that exporting is associated with plant size. However,
firms entering the export market are unlikely to substantially raise their productivity even
if they export continuously.
5
Melitz (2003) builds upon Krugman (1980) analyses of trade in the presence of product differentiation, increasing returns and monopolistic competition by incorporating firm level productivity differences. Also, he adapts the work of Hopenhayn (1992a, 1992b) to explain the endogenous selection of heterogenous firms in an industry. He showed how increases in industry’s exposure to trade (driven by trade liberalisation or the addition of new trading partners) lead to additional inter-firm reallocation towards more productive firms. Melitz (2003) thus explains how trade can generate industry productivity growth without necessarily affecting intra-firm efficiency. He emphasises trade can contribute to forcing the least efficient firms to contract or exit while promoting the growth and success of the more efficient ones. In his conclusion, most efficient firms export and increase both their market share and profits. Some less efficient firms still export and increase their market share but incur a profit loss while some even less efficient firms remain in the industry but do not export and incur losses of both market share and profit. Eventually, the least efficient firms are driven out of the industry.
Following this line of research, Helpman et al., (2004) constructed a multi-country, multi sector general equilibrium framework that explains the decision of heterogeneous firms to serve foreign markets either through exports or local subsidiary sales (FDI). They derived testable empirical predictions based on both cost and the extent of firm level heterogeneity in that sector. The predictions were tested on 1994 data of US affiliate sales and US exports in 33 different countries and 52 sectors. They found that in equilibrium, only the more productive firms choose to serve the foreign markets and the most productive among this group further chooses to serve the overseas market via FDI. In their exploration of implications of individual firms’ decisions for aggregate export and FDI sales relative to the domestic and foreign market sizes, it was shown that firm level heterogeneity is an important determinant of relative export and FDI flows.
Melitz and Ottavianno (2007) develops a model of trade with firm heterogeneity in terms
of productivity and endogenous differences in the toughness of competition across
markets – in terms of number and average productivity of competing firms. They
incorporated endogenous mark-ups using linear demand system with horizontal product
differentiation. They analysed how these features vary across markets of different size
that are not perfectly integrated through trade using evidence for US establishments
across regions. Their findings are that market size and trade affect the toughness of
competition which then feeds back into the selection of heterogeneous producers and
6
exporters in that market. Also, aggregate productivity and average mark-ups thus respond to both size of a market and the extend of its integration through trade (larger, more integrated markets exhibit productivity and lower mark-ups).
Further, De Loecker (2007) used matched sampling techniques to analyse whether firms that start exporting become more productive, while controlling for the self-selection into exports markets. He based his analyses on micro data of Slovenian manufacturing firms operating in the period 1994-2000. He found that export entrants become more productive once they start exporting and that productivity gap between exporters and non-exporters increases further over time. Also, productivity gains are higher for firms exporting towards high income regions.
Wagner (2007) used a panel data from1995-2004 at plant level to analyse the relationship between exports and productivity for German firms and found a significant productivity differential of exporters compared to non-exporters. There was an indication of self- selection of more productive firms into export market for West German plants but not for East German plants. Moreover, export starters that have low productivity at starting time fail as a successful exporter in the years after the start, and only those that were more productive at starting time continue to export. Also, Kox and Romagosa (2010) confirmed that only the most productive firms participate in exports and foreign direct investment when they used the 1997-2005 data on Dutch manufacturing firms and establishments. They however did not find any evidence for learning-by-exporting. On the other hand, Alvarez and Lopez (2005) found evidence for the learning-by-exporting in the Chilean manufacturing industry; evident existed for only new entrants but not for firms that export continuously. They also found a strong evidence that self-selection is a conscious process, by which plants increase productivity with the purpose of becoming exporters. Younger plants in the Chilean manufacturing industry are more likely to export more than older plants. Their analysis covered 1990-1996.
Not only has firm level export been analysed in developed countries but also in
developing countries. Roberts and Tybout (1997) develops a dynamic discrete-choice
model that separates the roles of profit heterogeneity and sunk entry cost in explaining
plants exporting status. Their work is on Colombian manufacturing plants operating from
1981-1989. When they expressed each plant’s current exporting status as a function of its
previous exporting experience, and observable characteristics that affect its future profits
7
from exporting, they found sunk cost to be significant and prior export experience is shown to increase the probability of export by as much as 60% points.
Clerides et al., (1998) analysed the causal links between exporting and productivity using plant-level data from Colombia, Mexico, and Morocco. They found that relatively efficient firms become exporters, however, in most industries, firms’ cost are not affected by previous exporting activities. They concluded that the positive association between exporting and efficiency is explained by self-selection of the more efficient firms into the export market.
3. Stylised facts
Before turning to empirical evidence of this study, I provide some basic statistics and give insights into implemented policies over the sample period. I further investigate if exporters characteristics exist within the sample.
3.1 Data source
This study is based on data produced by the Regional Program on Enterprise Development (RPED, hereafter). The data covers twelve years, collected in seven rounds over the period 1991-2002. Round one through three are annual surveys organised by the World Bank. The data is a representative of Ghanaian manufacturing firms classified by sector and size categories. For each firm, RPED collects data on productions, value added, number of employees, firm earnings, foreign affiliations or ownership, investments, exports, and other firm characteristics.
The panel nature of the data allows firms to be followed over time. Firms with less than 6 workers are classified as micro firms, followed by small firms with 6 to 29 workers.
Medium and large firms have been classified as firms with 30 to 99 workers and 100 or
more workers, respectively. The data was collected in locations such as Accra, Kumasi,
Cape Coast and Takoradi. In instances where some firms drop out of the sample, new
firms with the same sampling criteria as in the initial sample are incorporated so that the
sample of the firm remains representative over time.
8
3.2 Export trend and policies
Table 1 illustrates the trend over the sample period for exporting firms in the manufacturing industry in Ghana. It is observed that exports in the manufacturing industry was at the lowest in 1992. It however increased and reached its peak in 1994/95 and since then has been at low levels. The changes in number of exporting firms, especially the export boom in 1994/95, could be attributed to policies in the 1990’s described below.
Table 1: Firm exports 1991-2002 in Ghana.
1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 Number of
exporting firms
18 15 23 187 187 38 44 51 48 30 32 32
Percentage of exporting firms
2.6 2.1 3.3 26.5 26.5 5.4 6.2 7.2 6.8 4.3 4.5 4.5
Source: Prepared by author using data by RPED.
In general, the manufacturing industry in Ghana was relatively underdeveloped accounting for 0.8% until Economic Recovery Programmes (ERP) was initiated in 1983 (Ackah and Aryeetey 2012). Two of the main objectives of ERP are: (i) the development of internationally competitive industrial sector with emphasis on local resource-based industries with the capacity for increased exports and efficient import substitution, and (ii) introduction of measures that would attract entrepreneurs and investors into all major sectors (Asante and Addo 1997; Ackah and Aryeetey 2012).
In the early 1990’s, a product development division was created within the Ghana Export
Promotion Council (GEPC) to identify new products and producers, organise exporters
into production association and provide information to entrepreneurs. During this period,
education programmes were organised for exporters and export facilitators. Also, solo
exhibitions for exporters in selected countries within the Economic Community of West
African States (ECOWAS) were organised to promote exports into the region. Further, in
1993, programmes were established under Ministry of Trade and Industry to enhance the
supply capabilities of exporters by assisting them with incentives. These programmes are
the USAID sponsored Trade and Investment Programme (TIP) and the Private Enterprise
9
and Export Development (PEED) initiative sponsored by the World Bank. TIP aimed to eliminate obstacles to export expansion to enhance accelerated export growth through creating an enabling environment for the promotion of exports and improving on low capacity of firms to export by providing institutional support for exporters. Also, the Export Development and Investment Fund (EDIF) was implemented to boost the financing available from banking sector to exporters. The sectors targeted for support were textiles, garments, wood, and food. EDIF supported by exporters insurance, re- financing and credit guarantees through designated financial institution.
4These incentives in 1993 and the reduction of import tax rates by 5 percentage points (Ackah and Aryeetey 2012), which increased production due to lower prices of imported raw materials could explain the increased in the number of exporters in the years 1994 and 1995.
However, in the late 1990’s, the manufacturing industry experienced a decline in growth, from 5% per annum to about 4.4% between 1996 -2004 (Ackah and Aryeetey 2012). This decline was attributed to increase in the crude oil prices, high domestic interest rate, significant depreciation in the value of the cedi and domestic energy crisis that occurred in Ghana in the year 1998 (Ackah and Aryeetey 2012).
3.3 Share of firms
The data set contains information for a sample of 3,564 firms. Out of this, 315 are micro firms, 884 are small firms and medium and large firms accounts for 514 and 1,851, respectively. Table 2 classifies firms by their choice of globalisation modes. It presents the share of each firm size among the total number surveyed for each firm size category.
From the table, Export only indicates firms that export but are not affiliated to Multinational Enterprise (MNE
5, hereafter). On the other hand, MNE only denotes firms with MNE affiliation that do not export their products. The share of firms under this category produce only to serve the local market in Ghana. The category named both export and MNE indicates firms who simultaneously export and are affiliated to MNE.
Finally, domestic indicates firms that have no affiliations to MNE and neither exports their goods. Although their category may suggest they are not globalised firms, some of the firms may, however, be linked with global economies through other channels, such
4 To know more about trade policies in Ghana, see Ackah and Aryeetey’s book on Globalisation, Trade and Poverty in Ghana.
5 MNE is used in this study as firms who have at least 20% foreign ownership.
10
as raw material import which is not being analysed in this study. The firms under this category likewise firms under the export only category is all Ghanaian owned firms. They constitute 77.4% of the sample size.
In Table 2, it is observed that micro and small firms have highest percentages, 74 and 70.1 respectively in firms that neither export nor engage in MNE (domestic). Findings of Helpman et al., (2004) suggests that low productivity firms stay in domestic market.
Given this empirical evidence, information in Table 2 may suggest that majority of the micro and small firms productivity may be low compared to that of medium and of large firms. Within large firms’ size category, it is observed that 6.4% firms export without having affiliation to MNE and on the other hand, 23.6% are affiliated to MNE but do not export. It can also be observed that majority of the large firm exporters are MNEs. As the literature suggests that the cost of entry for being MNE is larger than the cost of entry for export; as such only firms with high productivity can become MNE. This can give a preliminary suggestion that large firms in the Ghana manufacturing industry as well as medium firms, given their observed values are productive.
Overall, 63.7% of firms in the manufacturing industry are domestic.
6Figure 1 shows that in the Ghanian manufacturing industry, 13.7% of firms engage in export but have no affiliations to MNE. Also, the percentage share of firms who are affiliated to MNE, but do not export is 16.5 and 6.1% simultaneously export and are MNEs. From Figure 1, it can be explained that in the Ghanaian manufacturing industry, 80.2% (i.e., domestic plus MNE only) firms are non-exporters, they only produce to serve the local market in Ghana.
This implies that 19.8% of firms in the Ghanaian manufacturing industry actively engage in exports, showing that firm level exports are a rare activity as emphasised by Wagner (2007). Within the MNE category, 73.1% or about three-quarter of MNEs produce to serve the local market. The data shows also that 69.4% of the exporters’ category export without having MNE affiliation.
6 This can be calculated from Table 1 by dividing the sum of domestic by sum of total firms (2271/3564).
11 Table 2: Share of firms within each firm size.
Firm size Export Only MNE Only Both Export and MNE
Domestic Total number of firms Micro
5 ≥ 𝑤𝑜𝑟𝑘𝑒𝑟𝑠
60 (19)
17 (5.4)
5 (1.6)
233 (74)
315 (100) Small
6 ≤ 𝑤𝑜𝑟𝑘𝑒𝑟𝑠 ≤ 29
210 (23.8)
39 (4.4)
15 (1.7)
620 (70.1)
884 (100) Medium
30 ≤ 𝑤𝑜𝑟𝑘𝑒𝑟𝑠 ≤ 99 101 (19.6)
96 (18.7)
60 (11.7)
257 (50)
514 (100) Large
100 ≤ 𝑤𝑜𝑟𝑘𝑒𝑟𝑠
118 (6.4)
436 (23.6)
136 (7.3)
1161 (62.7)
1851 (100)
Notes: The top values are actual values and percentages are in parenthesis.
Source: Prepared by author using data by RPED.
Figure 1. The overall share of firms.
Source: Prepared by author using data by RPED.
Table 3 disaggregates firms into sectors. Several notable differences in the percentage
share of firms within and between sectors emerge from this table. First, the extent of
globalisation varies considerably across sectors. The textile sector has the lowest
percentage of domestic firms at just 15% followed by the machinery sector 25.9%. Also,
12
the textile sector has the highest percentage of MNE at 80, followed by Beverage, machinery, and chemical at percentages 60, 55.5 and 46.7, respectively. However, none of the above-mentioned sectors seemed to be the most globalised sector. The wood sector has 25.7% the highest percentage of firms that are MNE and simultaneously exports. The sector also has the highest percentage of firms (29%) that exports but have no affiliation with MNEs and records the third lowest (27.5%) of domestic firms. Values in Table 3 suggests there may be high productivity in the above-mentioned sectors.
On the other hand, the sectors with the highest percentages of domestic firms (74.9- 94.4%) are Small-Scale Resource Intensive, Bakery, Garment and Furniture. Among these sectors are two (Bakery and Small-Scale Resource Intensive) that have no affiliation with MNE which also makes them the least globalised sectors. Interestingly, all the sectors participate in export and it can be observed that more than half the number of firms within the wood sector exports (about 54.7%).
Disaggregating the data this way gives more insight into the manufacturing industry. For instance, Table 2 gives insight into how much MNEs are centred in large and medium firms whiles it is observed that few MNEs are in micro and small firms. Some sectors in Table 3 do not have any affiliation with MNEs, however, empirical evidence suggests MNEs bring into an economy technological advancement which increase productivity and growth (Yeaple 2005). The observed values may suggest that large and medium size firms are reaping these technological benefits while the remaining firm sizes are in lack.
Table 2 however shows that micro and small firms have interest in exporting. The
percentages that export 25.5% and 20.6% within small and micro firms is even higher
than the 13.7% of firms that export within large firms. These firm sizes can reap the varied
benefits in exporting such as gains for workers in the form of higher salary and etc found
by empirical research.
13
Table 3: Share of firms within sectors.
Sector Export Only
MNE Only
Both Export and MNE
Domestic Total number of firms
Bakery 35 (12.2)
0 0 253
(87.8)
288 (100) Garment 92
(14.5)
19 (3)
5 (0.8)
520 (81.7)
636 (100) Textile 6
(5)
83 (69.2)
13 (10.8)
18 (15)
120 (100) Wood 80
(29)
49 (17.8)
71 (25.7)
76 (27.5)
276 (100) Furniture 91
(14)
48 (7.4)
24 (3.7)
485 (74.9)
648 (100) Metal 85
(13.4)
142 (22.3)
38 (6)
371 (58.3)
636 (100) Machinery 20
(18.5)
52 (48.1)
8 (7.4)
28 (25.9)
108 (100) Chemical 6
(3.3)
66 (36.7)
18 (10)
90 (50)
180 (100) Food 68
(14.9)
100 (21.9)
32 (7)
256 (56.2)
456 (100) Beverage 2
(3.3)
27 (45)
9 (15)
22 (36.7)
60 (100) Small-Scale
Resource Intensive
2 (5.6)
0 0 34
(94.4)
36 (100)
Notes: The top values are actual values and percentages are in parenthesis.
Source: Prepared by author using data by RPED.
In terms of share of sample according to location, Accra has a larger share of the sample size of 59%, followed by Kumasi with 31%, then Takoradi and Cape Coast receiving 6%
and 4% respectively. 43% of firms sampled from Takoradi actively engage in exports whiles 18% of firms sampled from Accra engage in exports. The samples from Kumasi and Cape Cost have the percentage of 20 and 19 respectively engaged in export activities.
However, with respect to the share of the total of 705 exporters, Accra received the larger
share of exporters of 52%. Appendix A depicts information on share of firms according
to location which can help policy makers in regional development.
14
3.4 How do exporters differ?
As found by previous studies, the characteristics of exporters such as wages and employment are higher than that of non-exporters, I investigate if such characteristics exist in the Ghanaian manufacturing industry. Following Bernard and Jensen (1995) and Alvarez and Lopez (2005), I perform ordinary least squares (OLS) regression on equation (a) to explain the characteristics of exporting firms.
𝑙𝑛 𝑥
𝑖𝑡= 𝛼 + 𝛽𝐸𝑆
𝑖𝑡+ 𝛾𝑍 + 𝛿
𝑠+ 𝛿
𝑡+ 𝜀
𝑖𝑡(a), where 𝑥
𝑖𝑡is the firm characteristic under test (e.g., number of employees, average wages, and capital intensity) in year t. 𝐸𝑆
𝑖𝑡is a dummy equal to 1 if firm 𝑖 exported at year t and 0 otherwise. Z is a vector of firm characteristics such as firm size and foreign ownership (MNE), δ
𝑠𝑎𝑛𝑑 𝛿
𝑡are sector dummy and year dummy which control for time shocks, while 𝜀
𝑖𝑡is error term. The parameter 𝛽 is what is of interest in this section as, it captures the difference between exporters and non-exporters. I therefore report only the estimated parameter value for the export dummy. The extended version of Table 4 can be found in Appendix B.
Using pooled OLS in this section seems advantageous. The reason is that the panel nature of the data allows to remove fixed firm effects and estimate the difference in the characteristics under test when a firm moves from producing entirely for domestic consumption to exporting some of its production. Moreover, when I have controlled for sector, firm size, and year effects, I am able to know that the firm characteristics hold over time, and across size class and are even true within micro sectors.
Table 4 shows the importance of exporting firms in the manufacturing industry in terms of employment (number of employees), average wages, and capital intensity (physical capital per worker). Exporters constitute only 19.8% of the sample but they show importance in the domestic economy. Like findings by Bernard and Jensen (1995, 1999), Greenaway and Kneller (2004), and Alvarez and Lopez (2005), and many other studies from different countries, exporter firms have superior characteristics.
After I controlled for firm size, technological innovation (MNE as a proxy), sector and year dummies, the estimated results are all positive and significantly different from zero.
I observe that on average, exporters employ about 5.9% more workers than non-exporters.
Also, exporters pay relatively more to workers than non-exporters, the wage differential
15
is approximately 42%. Another important difference between exporters and non- exporters concerns capital per worker; exporters are about 23.6% more capital intensive than non-exporters. Although the estimated result in this section confirms the characteristics of exporting firms, they do not give empirical evidence on whether firms self-select into export markets based on ex-ante productivity which is the objective of this study.
Table 4: Pooled OLS regression of firm characteristics on export dummy.
Dependent variable Export dummy Number of observations
R-squared Number of employed
persons
0.059 (1.97)*
1919 0.89
Wages 0.420 (5.92)***
1676 0.57
Capital intensity 0.236 (2.32)*
1879 0.45
NOTES: All dependentvariables
(𝑥
𝑖𝑡)
are in log. values of t statistics are presented in parenthesis.Variables that are significant are denoted by ***, **, and * at 1, 5 and 10 percent significant levels, respectively. Extended version of Table 4 is in appendix B.
Source: Prepared by author using data by RPED.
In this study, I include all firms for the purpose of analysing the relationship between export participation and productivity. The variable export is a dummy variable taking values 1 or 0 indicating whether the firm is an exporter or non-exporter, respectively.
Regarding productivity, it is measured as labour productivity defined in the data as value
added per worker. Bartelsman and Doms (2000) point to the fact that heterogeneity in
labour productivity has been found to accompanied by similar heterogeneity in total factor
productivity in the reviewed research where both concepts are measured. Chad Syverson
(2011) argues that high productivity producers will tend to look efficient regardless of the
specific way that their productivity is measured. Based on this argument and the fact that
it has been used by other literature (see e.g., Bernard 1995; Clerides et al., 1998; Vogel
16
and Wagner 2011; Kox and Romagosa 2010), I see labour productivity as an appropriate measure of productivity in this study.
4. Methodology
The self-selection hypothesis suggests that a positive relationship between firm performance and exports exists because only the most productive firms are capable of entering international markets, and that the level of competition abroad is higher than in the domestic market. In this context exporting will be profitable only for the most productive firms. Under this hypothesis, initial performance would be important to explain why some firms export and others sell only to domestic market. This study adapts Alvarez and Lopez (2005) and Kox and Romagosa (2010) model of export participation decision in which their results show that Chilean and Dutch exporting firms are more productive, larger and pay higher wages than non-exporting firms, respectively.
4.1 The probability of becoming an exporter
The main prediction of the heterogeneous firms trade model by Melitz (2003) is that firms opt for exporting if their productivity is sufficient to absorb the fixed entry cost in the export market. Following Kox and Romagosa (2010), I assume that actual export behaviour can be adequately described by a latent variable model in which the preference of firm ί in year t for exporting 𝑦
ί𝑡∗precedes actual exporting. The heterogeneous firms trade model can be reinterpreted in the following way. The decision to export 𝑦
𝑖𝑡∗depends on a set of observable firm characteristics 𝑥
𝑖𝑡and on an unobservable characteristic 𝜀
𝑖𝑡(e.g., the sunk entry costs firms expect to face in the export market).
In this study, the observable firm characteristic in 𝑥
𝑖𝑡is labour productivity (value added per worker). The assumed distribution of the unobserved characteristics 𝜀
𝑖𝑡determines the eventual export decision. Following Kox and Romagosa (2010), I assume that firm’s preference for exporting 𝑦
𝑖𝑡 ∗∈ {1, 0} depends on a linear additive relationship between the vector of observed 𝑥
𝑖𝑡characteristics and the unobserved 𝜀
𝑖𝑡characteristics that determine net export benefits.
𝑦
𝑖𝑡∗= 𝛽𝑥
𝑖𝑡+ 𝜀
𝑖𝑡(1)
17
If the latent decision variable 𝑦
𝑖𝑡∗exceedes a certain threshold level, which can be set to zero without loss of generality; then it is assumed that the firm exports. Consequently, if 𝐸𝑆
𝑖𝑡∈ {1, 0} is firm ί’s export status in year t, then what is observed is 𝐸𝑆
𝑖𝑡= 1 𝑖𝑓 𝑦
𝑖𝑡∗>
0 𝑎𝑛𝑑 𝐸𝑆
𝑖𝑡= 0 otherwise. Hence, the probability of exporting is formulated as follows;
𝑃(𝐸𝑆
𝑖𝑡= 1) = 𝑃(𝑦
𝑖𝑡∗> 0) = 𝑃(𝛽𝑥
𝑖𝑡+ 𝜀
𝑖𝑡> 0} = 𝑃(−𝜀
𝑖𝑡< 𝛽𝑥
𝑖𝑡) = 𝐹(𝛽𝑥
𝑖𝑡)
(2) where F denotes the distribution function of −𝜀
𝑖𝑡.Thus, equation 2 obtains a binary choice model that depends on the distribution of 𝜀
𝑖𝑡. As the scale of the firm preference 𝑦
𝑖𝑡∗is not identified, a normalisation on the distribution of 𝜀
𝑖𝑡is required
7.
Using a standard normal distribution, the binomial probit model of export decision is given by;
𝑦
𝑖𝑡∗= 𝛽𝑥
𝑖𝑡+ 𝜀
𝑖𝑡with 𝜀
𝑖𝑡~𝑁𝐼𝐷(0,1)
(3)
𝑎𝑛𝑑 { 𝑦
𝑖𝑡= 1 𝑖𝑓 𝑦
𝑖𝑡 ∗> 0 𝑦
𝑖𝑡= 0 𝑖𝑓 𝑦
𝑖𝑡∗≤ 0
To deal with the main objective of this study, investigating whether more productive firms in the Ghanaian manufacturing industry are more likely than less productive firms to start to export, I test the hypothesis by assessing the pre-export performance difference of export starters and non-exporters. According to the heterogeneous firm’s theory, a firm self-selects into export participation based on its relative performance in the domestic market. This implies that even before export starts there should be a positive performance premium
8. When adding control variables that may affect export participation decisions, the probit model is extended as in equation 4. In this instance, I consider years for firms not exporting in the first year and look at how the probability of beginning to export in the second year is affected by firm characteristics in the first year. The probit model is written
𝑃{𝐸𝑆
𝑖,𝑡= 1|𝐸𝑆
𝑖,𝑡−1= 0} = 𝐹(𝛽𝑥
𝑖,𝑡−1+ 𝛾𝐺
𝑖,𝑡−1+ 𝜆𝑅
𝑖+ 𝜂𝑇
𝑡+ 𝜀
𝑖𝑡)
(4) where 𝐸𝑆
𝑖,𝑡is a dummy variable equal to one if firm ί exported in year t, and 𝑥
𝑖,𝑡−1is labour productivity of firm ί in year t-1 while β is the parameter that reflect the impact
7 Usually this means that its variance is fixed at a given value Verbeek (2004). Since 𝐹(𝛽𝑥𝑖𝑡) is also bounded between 0 and 1, it is plausible to choose a standard normal distribution ∅(𝛽𝑥𝑖𝑡).
8 The premium is the productivity difference between exporters and non-exporters.
18
changes in productivity. 𝐺
𝑖,𝑡−1is a vector of firm characteristics and γ captures impact changes of the variables in the vector. The vector includes a MNE dummy that takes the value of one if the firm in year t-1 was affiliated to foreign ownership, and zero otherwise, as well as lagged (t-1) indicators of the firm’s human capital (using firm’s proportion of workers that completed secondary education as a proxy), firm age and three dummy variables for firm size (small, medium, and large). Previous literature suggests that these characteristics affect the probability of exporting. Concerning the variable MNE, the intuition behind is that firms with foreign ownership bring on board improved technologies and management skills that can translate into efficiency and productivity
9. Hence MNE is used in this study as technological innovation which may affect a firm’s decision to export. 𝑅
𝑖is dummy for sector that firms belong to. Finally, 𝑇
𝑡is a vector of year dummy to control for time shocks.
Previous studies have mainly used two estimation strategies for this binary choice model with unobserved heterogeneity. These are probit model with random effects and linear probability model with fixed effects. According to Bernard and Wagner (2001), the use of random effects requires that firm characteristics be uncorrelated with the regressors.
They emphasised that the required assumption is likely to be violated in export decision model as firm characteristics such as size and ownership characteristics are likely to be correlated with managerial abilities and unobserved firm effects. Several studies (see e.g., Roberts and Tybout 1997; Bernard and Wagner 2001; Alvarez and Lopez 2005; Kox and Romagosa 2010), however, use random effects probit specification in their analysis.
Bernard and Wagner (2001) who used both estimation techniques emphasised that the linear probability with fixed effects produces biases and inconsistent parameter estimates.
Bernard and Jensen (1997), who also used fixed-effects linear probability model, suggested the use of lagged variables as instruments. In accordance, they estimate the model in first differences, using two and three years lagged variables as instruments.
Since the use of two or three years lagged variables as instruments reduces the number of observations in my sample size to only a few observations, I employ the use of random effects probit regression in this study.
9 Helpman et al., (2004), extend the heterogeneous firm analysis to include MNEs and suggest that the sunk costs required to become a MNE are higher than that which is required to become an exporter. As such MNEs are more productive than purely exporting firms.
19
5. Results and Analysis
Table 5 present estimation marginal effects for equation (4) in column 3 and for two more restrictive models in columns 1 and 2. The model presented in column 1 include only MNE, human capital and firm’s age in addition to the main explanatory variable, labour productivity. In the model in column 2, I also include firm size and sector specific effects.
I further include year specific effects in column 3. As emphasised in the literature, a firm self-selects into export participation on the basis of its relative performance in the domestic market. As such, a firm’s initial productivity would be important to explain why some firms export and others sell only to the domestic market. This implies that in the findings there should be a positive productivity premium even before export starts. In estimating the results in Table 5, I use the past productivity (value added per worker) of firms as observed in RPED data set instead of the present productivity.
Table 5: Random effects probit model of export participation.
1 2 3
Labour Productivity
t-10.046 (3.49)***
0.045 (3.34)**
0.033 (2.14)*
MNE
t-10.147 (3.33)**
0.029 (0.64)
0.066 (1.29) Human capital
t-10.222
(3.29)**
0.125 (1.74)*
0.149 (1.80)*
Age
t-1-0.043 (-2.13)*
-0.051 (-2.48)*
-0.082 (-3.00)**
Firm size dummy No
Small
t-10.027
(0.52)
0.063 (1.05)
Medium
t-1-0.012
(-0.19)
-0.010 (-0.15)
Large
t-10.269
(3.91)***
0.261 (4.36)***
Sector dummy No
Bakery 0.044
(0.31)
-0.049 (-0.34)
Garment 0.230
(1.69)*
0.096
(0.79)
20
Textile 0.261
(1.69)*
0.192 (1.74)*
Wood 0.588
(8.89)***
0.313 (4.90)***
Furniture 0.167
(1.24)
0.062 (0.51)
Metal 0.148
(1.10)
0.040 (0.33)
Machine 0.242
(1.67)*
0.142 (1.17)
Chemical 0.306
(1.97)*
0.196 (1.78)*
Food 0.140
(1.03)
0.062 (0.52) Small-Scale
Resource Intensive
0.083 (0.28)
-0.759 (-0.11)
Year dummy No No Yes
Number of observations
1360 1311 1311
Log likelihood -841.74045 -750.64478 -447.87448
Notes: Probit regression with random effects. The marginal effects are calculated at the mean of each right-hand side variable. Values of z-statistics are presented in parenthesis. Variables that are significant are denoted by ***, ** and *, at 1, 5 and 10 percent, respectively. All independent variables except MNE, sector and year dummies are in log. Years are not reported in the table.
Source: Prepared by author using data by RPED.
A clear result is that the probability of exporting depends positively on the ex-ante labour productivity confirming the prediction of the self-selection model. The estimates in column 1 show that those firms that initially are more productive are more likely to enter export markets. A 1% increase in productivity increases the probability of beginning to export by about 4.6% in column 1. Through all three columns the effects remain positive and significant. This result is in line with the literature and findings from Alvarez and Lopez (2005) and Kox and Romagosa (2010).
The marginal effects, though positive and significant, declines through columns 1 through
3, from 4.6% to about 3.3%. It is observed in column 3 that increasing labour productivity
by 1 unit raises the probability of exporting by approximately 3.3%. The decline in the
21
coefficient could probably be explained by year shocks. Activities that occurred over the years could have influenced firm’s productivity (see Section 3.1). Column 3 shows that without controlling for year effects, results could be misleading.
With respect to MNE and human capital, results in column 1 shows that the decision to export is positively related to MNE and human capital and the estimates are significantly different from zero. According to the estimated marginal effects at the mean firms that are part of a MNE are 14.7% more likely to enter into the export market than firms that are not affiliated to MNEs. The effect of human capital remains significant in column 3 showing that increasing human capital by 1 unit raises the probability of exporting by 14.9%. Unlike Kox and Romagosa (2010), who used wages as a proxy for human capital, I use firm’s proportion of workers that completed secondary education as a proxy. The reason is in Ghana, most people who complete secondary technical and vocational education are employed at manufacturing industry. Involving such group into this study reflects the importance of skills acquired through education. The significance of the effect of MNE disappears in columns 2 and 3. The results however show that I cannot rule out that MNE increases the probability of starting to export, all else equal.
Interestingly, the effect of firm age is negative. Some studies have found positive whiles others, especially in developing countries, have found negative (see e.g., Alvarez and Lopez 2005). In this study it is estimated that younger firms are more likely to begin to export and the estimate increases in column 1 through 3, from 4.3% to about 8.2% with a higher significance level than that of columns 1 and 2. Since younger firms are more likely to export than older firms it could suggest that exporters may be firms that started operations with export market in mind. This result fits into Wagner’s (2007) theory of forward-looking firms. Here young firms have the desire to export tomorrow, and this induces their productivity today to be competitive on the export market.
The results also show that relative to micro firms the probability of a firm belonging to
the large size group to starting to export is significantly higher. In column 3 it shows that
large firms are 26.1% more likely to start exporting than micro firms. This is in line with
findings for other countries. The effects of sectors such as garment, textile, wood,
machine, and chemical are all positive and significantly different from zero in column 2,
however the significance level disappears for these sectors except for textile, wood and
22
chemical after year effects have been controlled for. These sectors are 19.2%, 31.3% and 19.6% respectively more likely to export than the beverage sector.
In general, column 3 shows that without controlling for year effects the results could be misleading. Also, because of the potential bias of the random effects estimator, I have as a robustness analysis estimated the model with the probit population average (GEE population-averaged model). As can be seen in Appendix C, this estimator gives similar results.
6. Conclusion
There are several empirical works that analyses the positive relationship between exports and productivity. This study has focused on whether more productive firms are more likely than less productive firms to start to export. Data on Ghanaian manufacturing firms from 1991 – 2002 produced by the Regional Program on Enterprise Development is used for this study. The study has also explored the significant differences in firms’
characteristics between exporters and non-exporters.
First, I have shown that firm characteristics that exist in the Ghanian manufacturing industry are consistent with evidence for other countries; exporters are superior in terms of number of employees, wages, and capital intensity. On the main objective of this study, I carry out empirical test in line with previous studies on whether firms self-select into export market based on their initial productivity using value added per worker as a proxy.
I find that firms that initially are more productive are more likely to enter export markets.
While the result indicate that initial productivity is a significant source for firms to enter export market, other observed firms’ characteristics also contribute to firms export decision. Firms that are large, young and firms that have a higher share or workers that completed secondary education are all more likely to export. However, Multinational Enterprises (firms with foreign ownership), which is used as a proxy for technological innovation, is not found to be significantly more likely to starting to export when sector specific effects are controlled for.
The evident that young firms are more likely to export may suggest that firms started
operations with export market in mind. If the firms actively target the export market, then
there could be policies to facilitate this process, since higher returns available in the export
23
markets may give incentives to increase productivity. A greater understanding of who exports should allow policy makers to focus resources better and concentrate their endeavours on strengthening the capabilities of exporters and motivating non-exporters.
Policies can be designed to increase domestic firms access to world market. Due to the evident that exporters employ more workers and pay higher wages, these firms can serve as mechanism for creating employment or reducing unemployment as a means of alleviating poverty and to achieve sustainable growth in Ghana.
Although my findings are to some extent in line with previous studies, this research has limitations. This result does not necessarily determine a unique causal relationship between productivity and exports. Even though my results show that exporters are more productive than non-exporters already before starting to export, it does not rule out that firm productivity can improve further upon entrance into export markets (learning-by- exporting hypothesis). Whether exporting improves productivity for Ghanaian manufacturing firms is an interesting topic for future research. Future research could also study which factors that affect firms decision to stop exporting, and how this can be prevented.
Acknowledgement
This study would not have been possible without the support of the Swedish Institute through their Scholarship for Global Professionals (SISGP). I am grateful to the Swedish Institute for their financial support throughout my study period at Umeå University.
I would like to express my appreciation to my supervisor David Granlund, for his
constructive guidance during the planning and development of this study. I also
appreciate the Regional Program of Enterprise Development for making data available
for free downloads to researchers. I would like to extend my thanks to my family for their
support and encouragement throughout my studies.
24
References
Ackah, C., & Aryeetey, E. (Eds.). (2012). Globalization, trade and poverty in Ghana.
IDRC.
Aitken, B., Hanson, G. H., & Harrison, A. E. (1997). Spillovers, foreign investment, and export behavior. Journal of International economics, 43(1-2), 103-132.
Alvarez, R., & Lopez, R. A. (2005). Exporting and performance: evidence from Chilean plants. Canadian Journal of Economics/Revue canadienne d'économique, 38(4), 1384- 1400.
Arnold, J. M., & Hussinger, K. (2010). Exports versus FDI in German manufacturing:
firm performance and participation in international markets. Review of International Economics, 18(4), 595-606.
Aw, B. Y., & Hwang, A. R. M. (1995). Productivity and the export market: A firm-level analysis. Journal of development economics, 47(2), 313-332.
Baldwin, J. R., & Gu, W. (2003). Export‐market participation and productivity performance in Canadian manufacturing. Canadian Journal of Economics/Revue canadienne d'économique, 36(3), 634-657.
Bernard, A. B., & Jensen, J. B. (1999). Exceptional exporter performance: cause, effect, or both?. Journal of international economics, 47(1), 1-25.
Bernard, A. B., & Jensen, J. B. (1999). Exporting and productivity (No. w7135). National bureau of economic research.
Bernard, A. B., & Wagner, J. (2001). Export entry and exit by German firms. Weltwirtschaftliches Archiv, 137(1), 105-123.
Bernard, A. B., Jensen, J. B., & Lawrence, R. Z. (1995). Exporters, jobs, and wages in US manufacturing: 1976-1987. Brookings papers on economic activity.
Microeconomics, 1995, 67-119.
Bernard, A. B., Jensen, J. B., & Schott, P. K. (2006). Trade costs, firms and
productivity. Journal of monetary Economics, 53(5), 917-937.
25
Bijsterbosch, M., & Kolasa, M. (2010). FDI and productivity convergence in Central and Eastern Europe: an industry-level investigation. Review of World Economics, 145(4), 689-712.
Clerides, S. K., Lach, S., & Tybout, J. R. (1998). Is learning by exporting important?
Micro-dynamic evidence from Colombia, Mexico, and Morocco. The quarterly journal of economics, 113(3), 903-947.
Conconi, P., Sapir, A., & Zanardi, M. (2016). The internationalization process of firms:
From exports to FDI. Journal of International Economics, 99, 16-30.
De Loecker, J. (2007). Do exports generate higher productivity? Evidence from Slovenia. Journal of international economics, 73(1), 69-98.
Falvey, R., Foster, N., & Greenaway, D. (2004). Imports, exports, knowledge spillovers and growth. Economics Letters, 85(2), 209-213.
Greenaway, D., & Kneller, R. (2004). Exporting and productivity in the United Kingdom. Oxford Review of Economic Policy, 20(3), 358-371.
Greenaway, D., & Yu, Z. (2004). Firm-level interactions between exporting and productivity: Industry-specific evidence. Review of World Economics, 140(3), 376.
Helpman, E., Melitz, M. J., & Yeaple, S. R. (2004). Export versus FDI with heterogeneous firms. American economic review, 94(1), 300-316.
Keller, W., & Yeaple, S. R. (2009). Multinational enterprises, international trade, and productivity growth: firm-level evidence from the United States. The Review of Economics and Statistics, 91(4), 821-831.
Kimura, F., & Kiyota, K. (2006). Exports, FDI, and productivity: Dynamic evidence from Japanese firms. Review of World Economics, 142(4), 695-719.
Kox, H. L., & Rojas-Romagosa, H. (2010). Exports and productivity selection effects for Dutch firms. De Economist, 158(3), 295-322.
Lodefalk, M. (2014). The role of services for manufacturing firm exports. Review of
world Economics, 150(1), 59-82.
26
Máñez‐Castillejo, J. A., Rochina‐Barrachina, M. E., & Sanchis‐Llopis, J. A. (2010). Does firm Size Affect Self‐selection and Learning‐by‐exporting?. World Economy, 33(3), 315- 346.
Melitz, M. J. (2003). The impact of trade on intra‐industry reallocations and aggregate industry productivity. econometrica, 71(6), 1695-1725.
Melitz, M. J., & Ottaviano, G. I. (2008). Market size, trade, and productivity. The review of economic studies, 75(1), 295-316.
Melitz, M. J., & Redding, S. J. (2014). Heterogeneous firms and trade. Handbook of international economics, 4, 1-54.
Roberts, M. J., & Tybout, J. R. (1997). The decision to export in Colombia: An empirical model of entry with sunk costs. The American Economic Review, 545-564.
Tomiura, E. (2007). Foreign outsourcing, exporting, and FDI: A productivity comparison at the firm level. Journal of International Economics, 72(1), 113-127.
Verardi, V., & Wagner, J. (2011). Robust estimation of linear fixed effects panel data models with an application to the exporter productivity premium. Jahrbücher für Nationalökonomie und Statistik, 231(4), 546-557.
Vogel, A., & Wagner, J. (2021). Robust estimates of exporter productivity premia in German business services enterprises. In MICROECONOMETRIC STUDIES OF FIRMS’IMPORTS AND EXPORTS: Advanced Methods of Analysis and Evidence from German Enterprises (pp. 239-263).
Wagner, J. (2007). Exports and productivity: A survey of the evidence from firm‐level data. World Economy, 30(1), 60-82.
Yeaple, S. R. (2005). A simple model of firm heterogeneity, international trade, and
wages. Journal of international Economics, 65(1), 1-20.
27
Appendix
Appendix A: Share of firms by location.
Source: Prepared by author using data by RPED.
28
