Manufacturing export performance, macro-economic adjustment and the role of networks Hans Hoogeveen and Takawira Mumvuma 1.1 Introduction Zimbabwean manufacturing exports have been subject to large policy changes. Under UDI foreign exchange controls and trade restrictions created a large import substituting manufacturing sector while exports were restricted to those few countries that defied the UN sanctions. After independence most controls were kept until, in 1991, the country embarked upon trade and exchange rate liberalization. The easing of biases against exporting, the opening up of the country for imports and the realignment of the exchange rate was expected to lead to important changes in the way trade takes place in Zimbabwe. However, trade liberalization was partly reversed in the late 1990s. The justification for the trade and exchange rate liberalization reflects the expected efficiency gains that follow from the improved functioning of markets and is based on traditional trade theory. Traditional trade theory and its modern versions emphasize relative factor endowments (Ohlin, 1933); skills development, learning and technological competence (Nurkse, 1953; Hirschman, 1958; Vernon, 1966), sunk costs, scale economies and agglomeration advantages (Krugman, 1992; Dixit and Norman, 1980), transport costs and concentration advantages (Brainard, 1993) as important sources of an agent’s international competitive advantage. These advantages can be exploited more efficiently only if market distortions are eliminated. Trade theories normally assume that trading relationships are anonymous, i.e. that agents trade indiscriminately with any other agent. In fact agents often do not engage themselves in anonymous relationships, and networks may actually form another comparative advantage determining the firm’s decision to export. This network factor captures the relationships an exporting firm has with its local and international customers, suppliers, competitors, distributors, industry organizations and government. Networks enhance reciprocity, trust and reputation building (Nadvi, 1998; Lou Egan and Mody, 1990); encourage learning and create conditions 1 conducive for product quality improvement, technological upgrading and organizational capability building (Grabher, 1988); reduce entry costs and minimize exit costs (DeBresson et al., 1991). In short, up to a certain level belonging to international export networks, reduces transaction costs and enhances export competitiveness.1 In this paper, firm level and macro-economic variables are used to gain insight in the determinants of Zimbabwe’s manufacturing export performance during the reform period. It is shown that after an initial decline, exports increased during the adjustment period but that macroeconomic variables are unable to explain this export response. Certain firm specific characteristics do have explanatory power however: large firms and firms with low capital intensity do better in exporting while the presence of networks is also shown to enhance firm level export performance. It is also found that a considerable potential for increased exports is left unexploited. The paper is organized as follows. In sections 2 and 3 a review of the evolution of the country's trade policies, incentives structures and manufacturing export patterns before and during the economic reform period is presented. In section 4 the available firm level information is used to construct a more detailed picture of the export performance of enterprises during the period of economic adjustment. This is followed in section 5 by looking for some complementary statistical evidence from the regression of the firms' export response function using firm level Regional Program on Enterprise Development (RPED) data. Conclusions are presented in section 6. 1.2 Trade Policy Before the Economic Structural Adjustment Period The break-up of the federation in 1963 was followed by Smith's UDI in 1965. The international community imposed trade sanctions on what was then called Rhodesia. Under these circumstances, the Smith regime had no option but to adopt an importsubstitution strategy characterized by strict market controls and regulations. These controls were directly imposed on external trade, foreign exchange allocation, prices, wages, interest rates, investment and repatriation of profits. Trade was carried out with only a few countries which defied the United Nations trade sanctions, especially South Africa, Mozambique, and Portugal. Ironically the sanctions resulted in the promotion of a diversified manufacturing sector servicing mainly the captive domestic market. The industrial success that was attained during the UDI period cannot only be attributed to the latter policy measures. The sense of unity and solidarity between 2 government officials and industrialists provided the necessary impetus to diversify and to engage in sanction busting activities whilst at the same time permitting the government to implement its trade policies with minimal opposition. At the same time, the collusion between the government and private entrepreneurs created a noncompetitive environment in which inefficiencies were not corrected by market forces. After independence in 1980 the inherited protectionist and restrictive trade policy regime was largely left intact. In fact the import controls were further intensified and administered through the same UDI rigid foreign exchange allocation system. A 20% currency devaluation in December 1982 and the subsequent daily adjustment of the exchange rate that took place up to 1990, and a number of initiatives for duty drawbacks, rebates and export incentive schemes all helped to offset the anti-export bias for the manufacturing (Ratnayake, 1994). Figure 1: Exports of manufactures in US$ and as percentage of GDP 14% 1200 % GDP 1000 800 8% US$ 600 6% 4% 400 US$ 1995 1994 1993 1992 1991 1990 1989 1988 1987 1986 0 1984 0% 1983 200 1982 2% 1985 % GDP 10% US$ million 12% Source: calculated from IMF Balance of Payment Statistics, various issues (GDP data); World Bank, Zimbabwe: Achieving Shared Growth, Country Economic Memorandum (export data, 1982-1993) and World Bank, unpublished data, (export data 1990-1995). The combined effect of these policies was the modest supply response of Zimbabwe's manufacturing sector exporters during the first half of the 1980s. The trend in figure 1 above demonstrates this. The various export incentives that were put in place failed to vigorously stimulate exports. This is due to the fact that the foreign exchange allocation system itself greatly determined resource allocation while failing to correct the fundamental policy design flaws of the Zimbabwean economy. The former resulted in a very distorted, bureaucratic and uncompetitive economic system in which firms' production capacities were domestically oriented and inefficiency was pervasive due to lack of market based internal and external competition, the high prevalence of rent-seeking behavior and the existing monopolistic and oligopolistic 3 market and production structures. According to the information contained in a report produced by the Zimbabwe Monopolies Commission 69% of Zimbabwe's manufacturing produce originated in firms accounting for at least 80% of gross output of these products (Government of Zimbabwe, 1992). Another dimension that can be brought in to explain the poor export supply response is the fact that the designers of the country’s export incentive system failed to make sure that the same incentives also benefited indirect exporters2. The shortage of foreign exchange became the greatest constraint facing existing and potential exporters. The sporadic export growth that took place during the last half of the 1980s can not be attributed to the export incentives per se that were put in place, but was a result of a reaction to the decline in effective domestic demand as well as the cross subsidization that was going on between the export and domestic markets in which exports were sold at a loss in order to earn the much needed foreign exchange (Mafusire, 1998). The losses were offset by profits generated on the captive domestic market. Therefore, what can be concluded here is that during the latter period, the Zimbabwean exporters of manufactured goods turned to the export market not as a deliberate strategy to promote their exporting capacities. It was only a way of defending themselves against declining domestic demand and the need to bridge the foreign currency shortage problem. Firms could only increase their capacity utilization levels, if they were in a position to export. Then they could use the same export receipts to import their raw materials, essential intermediate inputs and in some instances new machinery. The overall effect of these export incentives was that they engendered a high social cost on the rest of the economy, especially for domestic consumers and non-exporting firms who did not have free access to export incentives and therefore could not import whatever they wanted. Particularly with respect to the latter, they had to face low capacity utilization levels due to poor demand conditions, lack of raw materials and spare parts. The only opening that was available was to purchase these inputs domestically at prices far above the prevailing world prices. 1. 3 Trade Policy During the Economic Structural Adjustment Period With the start of ESAP, all the pre-reform export incentive schemes were systematically being phased out, but in their place the Export Retention Scheme (ERS) and the Inward Processing Rebate Scheme (IPRS) were introduced starting from October 1991 and during the second half of 1992 respectively. The introduction of the latter scheme was carried out in tandem with the expansion of goods that could 4 be imported without the foreign exchange rationing applying, namely goods that fell under the Open General Import Licence (OGIL)3. The ERS arrangement proved to be a very effective trade liberalization policy instrument (Gunning, 1996). Under ERS, exporters where initially allowed to retain 5 to 7.5% of their earned foreign exchange. Later this was increased to 25%, 50% and 100% in 1992, 1993 and 1994 respectively. Starting in January 1992 exporters were free to use their ERS allocations to import any goods of their choice with the exception of those which were found on the negative list4. Exporters could also trade their allocations at a premium on the open inter-bank market if they so wished. The IPRS allowed exporters to avoid payment of duty on imported production inputs as long as they were destined for the manufacture of exports. During this period the Zimbabwean dollar was devalued twice, in November 1991 by 56% and January 1993 by 28%. Individual Foreign Currency Denominated Accounts were introduced in June 1993. On 1 January 1994, Foreign Currency Accounts (FCAs) and a two-tier or a dual exchange rate system were introduced and OGIL, ERS and EIS arrangements were then abolished. Exporters could retain 60% of their earnings from export earnings in their corporate FCAs or sell it on the inter-bank market. They were required to sell the other 40% to the Reserve Bank of Zimbabwe at the official exchange rate (Mlambo et al., 1998). This implied that during this period Zimbabwe was effectively operating a dual exchange rate system. According to Gunning (1996) the mandatory selling of ERS funds constituted an implicit tax to the exporter since the central bank acquired these funds at the official exchange rate which was much lower that the going market rate. Another shortcoming of the ERS arrangement, like the pre-reform export incentive structure again, was that it was only designed to benefit final exporters at the expense of other indirect exporters. After the unification of the exchange rate regime in July 1994, there was a de facto 100% export retention permission with all foreign exchange requirements, including those of the government, sourced in the market. The objective of the latter move was to make the Zimbabwean dollar fully convertible and to eliminate the implicit tax on exports. The systematic decline of the Dollar index during the liberalization period reflects a reversal of the anti-export bias previously created by the country's overvalued exchange rate (table 1). The Dollar index measures the degree to which the domestic prices of non tradable goods are above or below normal international levels controlling for the country’s income level.5 If the index is above 100 the export sector is at a disadvantage through overvaluation and conversely if the index is below 100, it 5 is at an advantage through undervaluation. The index can also be interpreted as a measure of any given economy’s degree of outward orientation. Table 1: Development in Zimbabwe’s Real Effective Exchange Rates (Dollar Index) Year Dollar Index 1991 1992 1993 1994 121.1 112.5 105.8 102.9 1995 95.6 source: Bigsten et al., 1998 The severe 1991/92 drought adversely affected the manufacturing sector's ability to export during the first years of ESAP. The drought led to shortages and huge price increases of locally produced raw materials such as cotton; worsened the already bad macroeconomic fundamentals leading to an increasing lack of working capital; resulted in power shortages through load shedding, water rationing and loss of export earnings from the agricultural sector, which naturally worsened foreign exchange scarcity problems6. Exports in US$ dropped after 1989, and started to recover only in 1993 (figure 1). Measured as a percentage of GDP this recovery already started in 1992. A slower than expected short-run export response can also be expected because the liberalization program leads to a production shift in favour of internationally competitive firms. This implies that uncompetitive firms are likely to withdraw from exporting and some even went bankrupt as they could not deal with the increased competition of cheap imports. In the initial phase of the reform program one could therefore expect a decline in exports when uncompetitive firms withdraw while competitive firms have to make additional investments to respond to the new opportunities. Only at a later stage, exports are expected to increase. Figure 1 confirms such a pattern. Also, the termination of all the export incentive schemes in 1994, could have forced those firms which were exporting for just the sake of earning their own desperately needed foreign exchange to pull out of the export market and orient their production fully towards the relatively profitable domestic market resulting in the observed fall in export growth as a percentage of GDP after 1994. It can also not be excluded that while market efficiency was enhanced, institutional weaknesses and inefficiencies still continued to persist. Although the national export promotion agency, Zimtrade was rated as performing relatively quite satisfactorily 6 (Biggs et al., 1996), the institution failed to satisfy demand on trade information. Trade networks and other informal sources of information remain important to overcome the current institutional gap which cannot be filled solely by one well performing institution with a total staff of less than 40 people. Finally, the macroeconomic environment was also very volatile and uncertainties were large: exporters were not sure as to whether the program would be continued or disrupted. The latter, proved to be true later on as the government broke its relations with the program's main backers, the World Bank and the IMF in 1995. In addition, the gradual phasing out of the export incentives to Zimbabwean exporters at a time when the country's major trading partners were becoming more protectionist and offering more incentives to their exporters eroded the program's credibility.7 All in all the Zimbabwean trade liberalization program did not lead to the generation of the much anticipated high export growth in manufactured goods. By 1995 the percentage of GDP exported exceeded only slightly the level attained at the onset of the liberalization in 1990 (figure 1). Nonetheless export growth can be discerned for three of the four subsectors under study as can be observed from table 2 below. In accordance with figure 1 these statistics indicate a jump in export levels between the years 1991 and 1995 for the food, wood and textiles subsectors. The story is different for the metal subsector which shows stagnating exports during the same period. The export performance of the first three subsectors can be taken as an indication that initially the trade liberalization program managed to induce resources shifts from the non-tradable to the tradable sector. Table 2: Export values in million Z$ 1990 by sector Food Wood Textiles Metal 1990 1991 1992 1993 1994 1995 322 306 214 387 580 509 25 27 42 60 90 98 279 355 446 475 485 451 1265 1192 1127 1180 1076 1264 Source: Calculated from United Nations International Trade Statistics, various issues 1. 4 Trade patterns based on the RPED micro evidence In the previous section it was suggested that the export response of the Zimbabwean firms to the economic liberalization remained below expectations. In this section we 7 use the available firm level information to construct a more detailed picture of the export performance of enterprises during the adjustment period. To this end the RPED sample is examined across sectors for the period 1991 to 1995, for firms which provided export and sales information for each of these five years. The objective is to assess the extent of changes in exporting patterns over time, by sector and by destination. For many firms, exports are zero. Especially because of the possibility of threshold effects it is useful to distinguish both between exporters and non-exporters, on the one hand, and between exporting firms at different sizes. Weighted data are presented so as to reflect the population of all Zimbabwean firms and not just the firms in the sample. Focusing first upon the value exported, figure 2 presents its index since 1990. The pattern of exports that is obtained from the survey data corresponds to that presented in table 18. After a decline in 1992, due to the drought probably, and a slight recovery in 1993, exports pick up in 1994 and 1995. Between 1991 and 1995 the index for all firms increased from 100 to 145 in 1995, implying an average annual firm level increase in exported value in real terms of 7.7%. The value exported by small firms (those employing up to 100 employees) increases most rapidly: it tripled between 1991 and 1995. For large firms (those employing more than 100 employees) it increased by approximately 6.4% each year. Nonetheless the most important contribution to Zimbabwe’s export receipts is generated by the large firms given the fact that the total value exported by these firms is about 50 to 60 times as much as that of the small firms. Figure 2: Movements in the index of value exported in Z$ 1990 350 Index (1991=100) 300 250 all firms 200 small firms 150 large firms 100 50 0 1991 1992 1993 1994 1995 source: survey data (weighted) The increase in the value exported between 1991 and 1995, is partly explained by the increase in the proportion of firms that export. In 1991 12% of the sampled firms were 8 exporting; by 1995 this had increased to 19% (figure 3). Most of this increase was realised in 1993 and 1994. Two plausible explanations to explain the latter phenomenon can be advanced. It might have to do with the finalisation of the liberalisation of the trade regime during this period. Interesting in this respect is that 94% of the non-exporters claimed not to have experienced an increase in competition from outside the country, while 60% of the exporters claimed exactly the opposite. But it is also possible that the sudden increase in the number of exporting firms is due to a recovery from the 1992 drought. It could be that the drought and the contractionary effects of the reform program itself led to a mini domestic recession which forced manufacturing firms to turn to the export market as a “temporary” adaptive survival strategy, or that the increase in competition from outside forced firms to seek their markets elsewhere. The overall percentage of firms exporting masks the fact that the probability of a firm being an exporter varies greatly with firm size and the fact that large firms were much more sluggish in their response than small firms. In 1995 nearly one out of every three large firms was an exporter against one out of every four at the onset of liberalisation. For small firms in the sample the percentage of exporting firms in 1995 is lower, but at 17% it is still considerable. This relatively high percentage of small exporting firms is partly due to the fact that (generally non-exporting) micro enterprises are not represented in the sample. The increase in the proportion of small exporting firms is remarkable as this went up from 11% in 1991 to 17% in 1995. Figure 3: Percentage of firms that are exporters Percent of exporting firms 0.35 0.3 0.25 all firms 0.2 small firms 0.15 large firms 0.1 0.05 0 1991 1992 1993 1994 1995 source: survey data (weighted) Table 3 summarizes the transition rates in the export status of firms. Each row describes a transition from (not) exporting to (not) exporting in the following year. The entries in the table are the proportion of firms. On average, each year 3% of the non-exporting firms becomes an exporter while 2% of the exporting firms ceased exporting the following year. Given that the number of non-exporting firms is four 9 times as large as the number of exporters these percentages imply that each year a substantial number of firms start exporting. Most firms (87%) do not change their exporting status between 1991 and 1995: 77% remained non-exporters, 10% continued to be an exporting firm and 13% made at least one switch. There are few differences in the transition rates from exporter to non-exporter and vice versa between large and small firms, while the percentage of continuous exporters was substantially higher for large firms than for small firms (22% versus 8%). Especially interesting are the figures for 1993-1994. They show that a substantial number of non-exporting firms became exporters (6.8%), while very few exporting firms (less than 1%) stopped being exporters the next year. This suggests that firms easily switch in and out of exporting, a process that could also have been hastened by the firms’ survival motive which we have already mentioned above. The large number of small and large firms that make the transition to exporting in 1993-1994 and the relatively small number of firms that quit exporting, suggest that Zimbabwean firms have no problems to break into export markets and remain there. This is remarkable since the empirical literature on exporting shows that sunk start-up costs to exporting matter (e.g. Roberts and Tybout, 1997). These start-up costs vary from the necessity to invest in product-quality upgrading, overcoming informational constraints due to a lack of exporting infrastructure in the form of trading companies or distribution agents, and provisions to comply with regulations in export markets.9 Table 3: Firm transition rates in the export market 1991 - 1995 19911992 19921993 19931994 19941995 All Exports => No exports 0.4 1.2 1.0 3.2 firms No exports => Exports 0.8 Exports => Exports 12.0 No exports => No 86.8 exports 0.4 11.6 86.8 6.8 11.0 81.2 4.4 14.6 77.8 Small firms Exports => No exports 0.5 No exports => Exports 0.6 Exports => Exports 10.1 No exports => No 88.8 1.0 0.2 9.7 89.1 0.8 6.7 9.1 83.4 3.6 4.8 12.2 79.4 2.9 2.0 0.6 exports Large Exports => No exports 0.0 10 firms No exports => Exports 2.0 Exports => Exports 25.1 No exports => No 72.9 exports 1.6 24.2 71.3 7.7 23.8 66.5 1.9 30.9 66.6 source: survey data (weighted) If start-up costs to exporting matter nevertheless, then one expects small firms to be more affected than large ones. A relevant question is of course what is large in this respect. For non-specialized items like textile, large implies production in huge quantities requiring many employees. But for specialized items, such a snooker tables (in the survey one of the world’s largest high quality snooker table manufacturer was included) large can be as few as 10 employees. If start-up costs differ between the industrialized markets and those of the neighboring countries, for instance because of differences in product requirements, availability of information, or business cultures, then one expects that small firms sell primarily in the regional market while large firms are able to penetrate into the overseas markets. Table 4 presents the percentages of small and large firms that access different export markets in 1995. About 70% of both small and large firms serve the markets of the PTA and other SADC member countries so that no differences in entry barriers appear to exist with respect to these markets. Interesting is the difference that occurs with respect to South Africa, which is more popular with small firms than with large firms. The reason is found in the prohibitive tariff barriers that South Africa imposed on Zimbabwean textiles in 1992 initially meant to protect its local industry from a flood of textile and clothing imports originating from the Far East. The removal by South Africa of customs duty tariff advantages previously accorded to certain Zimbabwean products including textiles, clothing, shoes and furniture had a severe effect on the volume of exports destined for this market since Zimbabwean exports became uncompetitive. As a consequence, only 44% of the large exporting textile firms sell to South Africa, as opposed to 76% of the small ones. Apparently these small firms due to the informal nature of their cross-border export arrangements have found effective ways to deal with these tariff barriers. As expected the European market is more easily accessed by large firms than by small firms, while rather surprisingly the differences for the United States between small and large firms are negligible. The inability of small firms to access the EU market can be explained by some of the more qualitative reasons that were given by some exporting firms in order to explain why they were not in a position to export to the EU market such as prohibitive transportation costs, poor product quality, lack of contacts and market 11 information, satisfying EU quality norms, lack of capacity to produce required quantities and antiquated technologies. Table 4: Percentage of exporting firms exporting in 1995 to: Small Large PTA or SADC countries South Africa Other Africa member 68 69 77 0 49 5 0 18 0 14 12 3 31 0 11 5 Middle East Europe Asia United States Other source: survey data (weighted) If we consider the actual proportion of output which is exported this percentage is low as can be observed from table 5. On average firms export less than 2% of their output. Remarkable is that this percentage decreases between 1991 and 1994 only to recover in 1995. There are, again, large differences between small and large firms. The percentage of output exported by small firms declines until 1995 in which year it nearly doubles. That of large firms increases steadily from year to year. Most firms do not export at all, so that on average only 2% of output is sold on foreign markets.10 Exporting firms sell approximately 15% of their output abroad. This percentage varies between small and large firms: small exporting firms sell about 12% of their output abroad, large firms more than a quarter of their production. Table 5: Percentage of output exported, all firms and exporting firms only 1991 1992 1993 1994 1995 All firms Small firms Large firms 1.7 2.5 0.5 1.3 1.7 0.6 1.1 1.2 0.9 1.0 1.0 1.1 All exporting firms 23.1 16.9 15.4 9.1 1.8 1.9 1.5 14.2 Small exporting firms 25.7 17.3 13.4 6.4 11.4 Large exporting firms 13.0 15.1 22.1 20.8 26.6 source: survey data (weighted) 12 Table 6 presents the value of exports by destination. Distinguished are destination of exports in value terms at the sectoral level and for all firms combined. The regional market is very important. Nearly 50% of the exported value measured as the average export intensity at the firm level is sold in South Africa or one of the other countries in the region. This is not surprising given the special trade agreements between Zimbabwe and its regional trading partners such as Botswana, Malawi and Zambia. The other half of exports is sold in the markets of the industrialize world. Europe (33%) is the most important industrialized export destination followed by the United States (15%). Interesting is the sectoral composition of the destination of exports. Food and metal products are primarily sold in the African market and very little is sold in Europe and the United States. That the food sector has difficulty to sell in these markets is probably due to the strict hygiene and health regulations that apply in these markets. Metal products, such as standardized products like barbed wire due to their bulky nature attract higher transportation costs and therefore are expensive to ship over large distances while custom made metal products such as fences require a close relation with the customer. Wood products and ready to assemble furniture especially do not have the latter disadvantages. It is primarily sold in Europe. Textiles are mostly sold both in Europe and the United states. The dominance of textile exports to the EU can be explained by the ACP preferential treatment enjoyed by Zimbabwean textile exporters on entering the EU market as long as they can satisfy the rules of origin. Textiles are also by far the most important exported product in value terms: 61% of the exported value are textiles of which less than one third is sold within Africa. Table 6: Destination of exported value in 1995 (measured as the average export intensity at the firm level) as percent of total firm exports. Food Wood Textile Metal Total PTA or SADC countries South Africa Other Africa Middle East Europe Asia United States Other Total member 15.1 0.1 6.6 5.3 27.1 3.2 0.0 0.0 5.4 0.0 10.7 0.2 0.0 26.5 0.0 2.4 0.0 0.0 0.0 0.0 19.7 0.3 1.3 32.8 0.0 0.5 0.1 0.3 0.1 21.4 8.9 13.8 3.4 61.3 0.7 0.0 8.4 15.0 3.9 100.0 3.3 0.1 1.3 0.9 0.0 13 source: survey data (weighted) At the sectoral level, as can be observed from the table 7, firms have been responsive to the changes in the relative prices in favor of exports. The wood and textile subsectors are remarkable in this respect, as the percentage of textile firms exporting doubled while the percentage of wood firms exporting more than tripled between 1991 and 1995. Nevertheless, the percentage of exporting firms is still the lowest in the wood subsector. Those wood firms that do export are generally large firms. For all sectors holds that a (much) larger proportion of the large firms are exporters than of the small firms. Table 7: Percentage of firms exporting by sector 199 1992 1993 1994 1995 1 All Firms Food Wood Textile 24 3 10 24 3 10 23 3 10 26 9 19 23 10 21 Metal 13 16 10 11 15 Small Firms Food Wood Textile Metal 22 0 8 10 22 0 8 14 22 0 8 8 24 5 18 6 21 5 20 10 Large Firms Food Wood 43 16 43 22 34 22 54 30 51 36 24 24 27 24 29 18 27 31 28 32 Textile Metal source: survey data (weighted) That networks play a role in the orientation of exports is suggested below. Table 8 shows that firms which are (partly) foreign owned export a larger fraction of their output: on average 3.2% between 1993 and 1995 against 1.5% for firms with Zimbabwean owners only (weighted figures). This also holds for firms that are a subsidiary of a multinational corporation. They export 5.9% on average against 1.2% for other firms. 14 Table 8: Percent of output exported by foreign ownership/subsidiaries of multinationals* Foreign owners Subsidiary of multinational 1993 1994 1995 average * yes 2.8 2.1 4.7 3.2 no 1.3 1.4 1.7 1.5 Yes 4.7 6.2 6.7 5.9 no 1.1 1.1 1.4 1.2 Included are firms with nonmissing exports for all years between 1993 and 1995 source: survey data (weighted) Cultural background may serve as a network variable. Table 9 seems to confirm this as it indicates that European owners are more internationally oriented and export especially to white-dominated markets of Europe, the United States and South Africa. The high penetration of South Africa might be explained by fact that during the period of sanctions, South African entrepreneurs kept dealing with their Zimbabwean counterparts, thus building a network of trust, while the exports to Europe and the United States could just as well be explained by the fact that for large firms, which are usually white owned and older , it is easier to penetrate these markets. A multivariate analysis would have to determine whether racial origin is instrumental for firm size or whether is represents other factors. That it is easier to trade with people of a similar cultural background is confirmed by the fact that African owners specialize in trading with customers in other African countries, while the comparative advantage of European owners appears to lay elsewhere. This leaves unexplained why Asians do not export to Asia, while they are over-represented in exports to Europe and the United States. Partially this has to do with the fact that Asians are concentrated in the textile business . This is an export oriented industry, which is able to export to Europe and much less to Asia where a similar industrial capacity exists. Also it could be argued that due to their long trade and entrepreneurship history, the Asians slowly managed to penetrate these European and American export networks either directly or indirectly. Table 9 : Ethnic distribution of firm ownership and destination of exports Exporting firms African Ethnic origin South of owner Africa 71% 7% Europe USA Other Asia Africa 12% 15 15% 37% 0% European Asian Number firms of 19% 10% 100% 135 76% 17% 100% 24 67% 85% 22% 0% 101% 100% 22 9 55% 9% 101% 27 0% 0% 0% 0 In summary, the increase in the number of exporting firms and in the value exported suggests that the policy reform in favour of exports does bear fruit. The firm level evidence sheds some new light on the trade issues raised using macro data. Discussions of the problems facing exporters in Africa frequently assume that the problem is enabling them to break into export markets. This appears to be less of a problem. Both large and small firms are in a position to switch into exporting, though there is some evidence that large firms have a comparative advantage in accessing the European market. The evidence suggests that networks are important. In that sense, the presence of a white minority in the manufacturing sector could be an asset in penetrating overseas markets. These are encouraging conclusions though two qualifications need to be made. First the percentage of output exported is low. The average exporting firm sells about 85% of its output on the domestic market, and few firms specialize in exports: as a matter of fact less than 5% of all exporting firms sold more than half of their output abroad. Secondly, in value terms large firms are by far the most important contributors to exports, but less than one out of three of them is actually an exporter. Hence there still is a considerable number firms who apparently are not interested or not able to be exporters. This is worrying because these large firms especially should be in a position to penetrate and operate in the international markets. 1.5 Determinants of Export Performance in the Manufacturing Sector In this section a statistical approach is taken to the question which firms export, how much they export and to which foreign market(s) they export. We examine which variables affected trade, measured both as the number of firms exporting and as the percentage of output exported. To do so macroeconomic variables, firm characteristics and network variables are incorporated in the analysis. Unlike in the previous section where five years of survey information could be used, in this section we limit ourselves to three years due to missing observations for the years 1991 and 1992 for most variables except exports and employment. Included are firms for which information on exports was obtained for all three years. 16 Suppose that the underlying continuous version of the model explaining the percentage of sales exported is given by: Yi* X i i where Y * is a latent variable indicating the optimal percentage of sales exported, X is a vector of explanatory variables reflecting the profitability of being an exporter, and is a normally distributed disturbance term. Because exports are bounded by zero from below, the actual observed percentage of sales exported Y is given by: Y Yi * for Yi * 0 Y 0 for Yi * 0 This model is also called censored regression model or Tobit model, and can be estimated directly by maximum likelihood or a 2-step estimation procedure using a Heckman sample selectivity correction term (Greene, 1993, ch. 22). A 2-step procedure is preferable, if there are some variables that differently affect the chances of being an exporter from the percentage of output exported. In our case we suggest that network variables are determinants of the likelihood that a firm is an exporter, but not of the percentage of output exported11. In this section we use therefore the Heckman 2-step procedure for estimating selection models. First we estimate a probit model by maximum likelihood for the probability that a firm exports, and second a linear regression model including a Heckman sample selectivity term (Mill's ratio) for the percentage of sales exported given that a firm has exports. The independent variables included in the estimations require some explanation. First, size may be important for several reasons. Exporting may involve the firm in higher marketing costs than domestic sales. The larger the firm the lower the average cost of exporting. Export orders may also require production in bulk, requiring a large firm to fulfill the orders, especially since subcontracting in Zimbabwe is an exception rather than the rule (Biggs, 1996). Also, larger firms, through their many contacts, are usually better in dealing with bureaucracy and so they are in a better position to access the prevailing export incentives. For example of all the firms that applied for an export subsidy under the export incentive scheme, 95% of the large firms and 75% of the small firms actually reported to have received it in 1992. Large firms waited 29 weeks against small firms who waited 37 weeks on average (source: survey data). Size increases access to banking services which may be more important for exports 17 than domestic sales, especially when exporters need to invest to upgrade their equipment. To capture the effect that firms which invested in efficiency and hence in (more specialized) equipment are more likely to export, we include the capital labor ratio to act as a proxy for productivity. Zimbabwean producers generally lack specialized equipment for making different types of goods. The country’s isolation typically led manufacturers into producing a large range of products, but in small quantities of each style for the local market. As a consequence firms have basic machines but no specialized equipment needed for specialized, volume oriented production12. Alternatively, it may be that exporting is concentrated in labor intensive activities given that Zimbabwe is relatively abundant in labor. Then export propensity falls with the capital-labor ratio and we expect a negative correlation between the capital-labor ratio and the ability to export. Capacity utilization is included to verify the relation between domestic sales and exports. If exporting is considered not to be as profitable and therefore less attractive than sales on the domestic market, then higher levels of capacity utilization in order to meet domestic demand have a negative effect on exports (unless additional investments are undertaken to increase production capacity). If exporting is regarded to be more profitable than domestic demand, then no effect of capacity utilization on exports is expected, as the firm will try to export as much as possible independently of the strength of domestic demand.13 A positive effect of capacity utilization on exports can be expected if a higher capacity utilization implies a higher degree of efficiency. Firm age is included to reflect the influence it may have on capital costs and to approximate the extent of a firm’s learning experience. The percentage exported may also depend on whether the firm requires imported inputs. Under the ERS scheme exporting became attractive because firms were allowed to keep (part of) their export earnings. Firms requiring imported inputs thus have a larger incentive to export. Sector dummies are included to capture the effect that the changes in the macro environment may affect sectors differently. Tradable sectors will benefit more than sectors producing non tradables. Other sectoral differences, in access to capital or dependence on weather may also exist. The sector dummies capture these effects. The Dollar variable indicating the abolishment of the overvalued exchange rate and the disappearance of the anti-export bias (table 1) is included to link macro policy to firm behavior and a positive relation between the Dollar variable and export 18 performance is expected. However this relation is not straightforward as the removal of the anti-export bias was accompanied by an increase in macroeconomic instability as fiscal discipline was not achieved. The causes are multiple and include the fiscal consequences of drought, the soaring public interest bill following the improper sequencing of the economic liberalization, especially with regard to the early liberalization of the financial sector14 and the powerful industrial, political and military interest groups laying continued claims on the budget. Whatever the reasons, the fact remains that the budget deficit was unsustainably high creating uncertainty about future market conditions. Telling in this respect is that the first year of the survey, in 1993, 18% of the exporting firms and 41% of the non-exporting firms mentioned uncertainty about government industrial policy as their most pressing problem. In such uncertain circumstances a wait and see attitude can be optimal (Pindyck, 1988 and Dixit, 1989) if increased exports require sunk investments in marketing or specialized equipment for instance. The rationale for such an attitude is that an investment expenditure involves the exercising or elimination of the option to productively invest at any time in the future: expenses for instance to establish contact in foreign markets eliminate the option not to expend now. In the face of uncertainty it might be optimal to keep options open and follow a wait and see approach, especially if the expenses are characterized by a high level of irreversibility. This reduces the firms ability to produce large export orders but brings some greater production flexibility at the firm level. Another way to keep options open is by not buying specialized equipment but machines and tools with a wide applicability. Calvo (1988) further illustrates this aspect by focusing on the importance of the lack of credibility on the continuity of the reform. He shows that this lead to a high purchase and storage of importable goods during the temporary liberalization. These effects are not just theoretical artifacts. Ncube et al. (1999) provide several examples of the existence of speculative behavior with respect to imports at various stages of the reform process. Using the RPED dataset they also find that the liberalization did not induce investments in the export sector. This despite the fact that in 1994 62% of the exporting firms indicated that the export opportunities had improved, while 84% said that the opportunities to obtain foreign equipment not available before the reform program had improved. Bigsten et al. (1998) who were the first to use Dollar variable in an analysis of trade performance also found mixed results. In an analysis of firms in four African countries: Cameroon, Ghana, Kenya and Zimbabwe, they found only for large firms in Cameroon a positive response to greater macroeconomic openness. The decision whether or not to export is assumed to depend on whether people encounter export opportunities through their own specific efforts or via their 19 networks. Such networks are not confined to family members and friends. Multinational companies and other private firms from abroad also offer such linkages though firms with foreign owners could also be less interested in exporting if firms were mainly set up to cater for the domestic market as part of their parent companies’ global production and marketing strategies. But links through foreign ownership are likely to provide the domestic exporting firm with the necessary information on export opportunities, technical know-how, learning possibilities, offshore financial resources, assistance with trouble shooting, marketing and technology information as well as human resource skills. Firms which are owned by private owners, whether Zimbabwean and foreign owners jointly or foreign owners alone are likely to have a different network and hence differential access to foreign markets. Networks can also be useful to enhance trust. Export orders entail a great deal of risk since the transacting parties are separated by country borders, social and legal systems, language barriers and as well as geographical and technological distances. Several formal and informal mechanisms have to be developed in order to deal with such contracting risks (Fafchamps, 1998). Letters of credit, are an example of a formal mechanism: it is an irrevocable promise of payment, providing the conditions of sales are met. Unfortunately, in Zimbabwe there are significant obstacles to using letters of credit. Biggs et al. (1996: 73) cite a German importer who does not use letters of credit in Zimbabwe, “because bank fees and bureaucracy are too costly”. In such an environment, trust between the contracting parties plays an important role. Having trust in the exporter and knowing him in person, will keep the foreign buyer fully informed of developments that affect him and will in turn allow a good understanding of the problems that may be affecting the exporter. A well trained sales staff who is aware of these requirements is thus a necessity to be successful as an exporter. In the regressions, this is approximated by the level of education of the sales staff. Location in a city also captures this if it is interpreted as a proxy for the presence of business enhancing infrastructure (telephone, airport, roads or golf courses). Another possibility is that communication (and hence trust) is enhanced between people of the same racial background (Cornell and Welch, 1996) so that African entrepreneurs might have a larger access to or opt for the regional market while entrepreneurs of a European background have easier access to the European market. Commonly shared norms and values may help in facilitating the easiness with which contracts are enforced and therefore generally go a long way in minimizing the overall ex ante and ex post transaction costs. Under these circumstances the decision to export and the destination of exports may be related to the racial origin of the 20 exporter. To deal with these factors a racial dummy (European) is included in the regressions. The use of race as a network variable calls for a careful interpretation as the possibility of omitted variable bias needs to be considered before a significant coefficient on ethnic origin is taken as evidence of networks and not of other form of discrimination (Heckman, 1998). The large number of variables included in the regressions, including those which are strongly correlated with ethnic origin such as firm age and firm size, and the absence of evidence of discrimination in the attribution of bank credit (Fafchamps, 1998a), makes us confident with the assumption that the racial dummy indeed measures network effects and not discrimination. The regression results are presented in table 1.10. The probability of being an exporter and the percentage of sales exported given that there are exports increase with firm size. Higher capital intensity reduces the likelihood that a firm will be an exporter, implying that firms which make use of Zimbabwe’s comparative advantage in labor endowments do export. This is supported by the claim by 75% of the exporting firms that they are competitive in the region because of their labor costs and not because of the availability of capital. As a matter of fact, only 17% indicated the availability of capital as their competitive edge over firms in the region. Being an exporter and capacity utilization are also positively related. If higher capacity utilization implies a higher degree of efficiency, the result that capacity utilization matters for the decision to export but not for the percentage of output exported supports the findings by Clerides et al. (1998). These authors report for a sample of Columbian, Mexican and Moroccan firms that relatively efficient firms become exporters and that the positive association between exporting and efficiency is explained by the self-selection of the more efficient firms into the export market. Firm age also explains the decision to export but is insignificant in the explanation of the percentage exported. The network variables are also significant. Firms with educated sales staff, that are part of a multinational company or that have European owners find it easier to become exporters. The significance of the educational level of sales staff shows the importance of those who are directly involved in nurturing exchange relationships between firms. It helps to identify where to put effort in terms of commercial export skills development. Interestingly foreign ownership, state ownership or being located in a city do not explain whether a firm is an exporter. The dummy variables indicate that firms in the wood sector have an advantage in being an exporter, while firms in this sector and in the textile sector export a larger share of their output. This confirms what was found in the previous section: metal and food firms produce less tradable goods than wood and textile firms. Finally, the Dollar variable is insignificant, so that 21 it is unclear what is the effect of the macroeconomic environment on the decision to export. Table 10: The percentage of sales exported, with Heckman correction for selectivity bias Decision to export Output exported Lagged log employment Log capital per employee 0.628 -0.275 4.71 -2.40 8.976 -1.314 5.94 -0.78 Log capacity utilization Log firm age Log years of education sales staff D-multinational D-European owner 0.536 0.618 1.475 1.254 0.516 1.78 2.92 1.81 1.81 2.33 -0.150 0.357 -0.03 0.12 D-imported raw materials D-state ownership D-foreign ownership 0.103 0.470 -0.183 0.27 1.05 -0.62 7.233 1.03 D-city D-textile D-food D-metal D-wood Dollar Constant 0.039 0.326 -0.194 0.512 1.807 0.024 -8.372 0.11 1.03 -0.56 1.17 2.31 0.97 -2.36 16.233 -4.844 4.807 46.618 -0.218 -13.835 11.708 3.50 -0.85 0.80 5.18 -0.58 -0.31 17.2 9 Observations 2(df) 258 169.4 (27) Lagged log capital per employee and capital per employee have a correlation coefficient of 0.96. The same holds for lagged log employment and log employment where the correlation coefficient is 0.97. To avoid the endogeneity problem lagged capital per employee and lagged capacity utilization are to be preferred. However the inclusion of lagged capital per employee generates an efficiency problem in that about one third of the sample is lost. The same holds for lagged capacity utilization. This problem does not occur for employment so that lagged employment and current capital per employee and capacity utilization are included. It was checked whether this makes any differences, and in fact it does not in the size of the coefficients but it does in the level of significance. Due to the larger sample size, capital per employee becomes significant. Several cross effects have been included in the estimations, but none of them was significant and they were subsequently dropped from the analysis. 22 With respect to the percent of output exported, firm size is a clear explanatory variable. Hence large firms not only are more likely to be exporters, they also export a larger share of their output. Other than firm size, none of the independent variables is significant except the dummies for firms in the textile and wood subsectors. Finally we present the logit regression on whether a firm is selling in the region or on the world market (i.e. Europe or the United States). In the regression for world, two variables had to be dropped due to high multicollinearity: imports of raw materials and food sector. If entry barriers are important then one may expect that firms that sell in Europe or the United States have characteristics that differ from those for firms that sell in the more nearby markets. The estimations support this convincingly. First, firm size is important to sell on the world market but not for sales in the region. This confirms the earlier suggestion that entry barriers are more important for the overseas markets than for the regional markets. In addition firms that sell on the world market do so because of a comparative advantage due to low labor costs, given that the capital per employee enters with a negative sign. Capacity utilization is no explanatory variable for the decision to export, unlike firm age. Firm age has a strong impact on regional as well as extra-regional exports. The same applies for personal contacts, given the fact that a European owner is an explanatory factor. Interesting is the observation that the level of skill of the sales staff matters for exporting to the region, but not for exporting on the world market. A possible explanation might be that overseas transactions – which we expect are of significant size – are administered primarily by the firm owner, while smaller export orders – usually in the region – are dealt with by the sales staff. If they are better trained, then this clearly raises the probability of a firm exporting. Table 11: Logit regression on whether a firm exports within the region or to the world market Region World Lagged log employment Log capita per employee Log capacity utilization Log firm age Log years of education sales staff 0.218 -0.052 -0.553 2.528 3.569 1.13 -0.27 -1.01 5.92 2.30 1.224 -0.991 1.061 2.706 -0.259 4.54 -3.77 1.41 4.13 -0.17 D-multinational D-European owner D-imported raw materials 1.054 1.251 -0.672 0.95 2.95 -1.00 5.757 2.312 4.18 3.52 23 D-state ownership D-foreign ownership D-city D-textile D-food D-metal D-wood D-1994 D-1995 0.393 -0.994 1.366 -2.275 -2.646 -1.031 -0.987 0.116 0.386 0.46 -2.09 1.91 -3.67 -3.91 -1.26 -0.37 0.28 0.86 -0.293 -3.868 -2.786 3.489 -0.28 -4.35 -3.43 4.68 2.612 10.870 0.197 0.780 2.45 5.11 0.36 1.35 Constant -17.362 -3.71 -6.491 -1.41 Observations 2 (df) 258 149.7(17) 258 172.0(15) Ownership matters for the direction of exports. As part of a multinational a firm is more likely to export to the world market, while this does not affect the probability for regional exports. It suggests that firms that are part of a multinational corporation are included in the mother company’s global production and marketing strategies. This does not hold if the firm is (partly) foreign owned (mainly by South Africans). In that case the probability of being an exporter is not enhanced. Finally, the sectoral dummy variables confirm our previous ideas. Food firms are at a disadvantage to sell in the regional market. This holds even stronger in the world market where this variable had to be dropped due to high collinearity. Hence the food sector is better classified as a non-tradable sector. Textiles have difficulty to be sold in the region. This is in line with the high duties raised on textile exports in the largest economy in the region: South Africa. Firms in the wood sector find a ready market in Europe. Unexpected and not easy to explain are the significant dummy for exports by metal firms to the European market and the significant but negative sign for being located in the city. 1.6 Conclusion and Summary In this paper we have attempted to use macro-economic and firm level variables to explain the modest supply response to trade reforms by Zimbabwean exporters of manufactured goods. Aggregated macro statistics and indicators show that exports managed to grow during the reform period, but at a rate that was below what was hoped for. The Dollar index, included in the regressions to reflect the changes in the macro economic environment was insignificant, suggesting that firms did not really 24 respond to the changes in the policy environment. This could very well be due to the uncertainty that accompanied the introduction of the measures that eliminated the anti-export bias. Networks appear to be important. Firms which are part of a multinational company or which are owned by a European owner and which have better educated sales staffs are more likely to be exporters. This is also confirmed by the fact that older firms are more likely to export. Other factors, more in line with traditional trade theory, are also significant. Firms that are large and which make use of the available labor endowments (textile and wood sector) do better in exports. Analysis of the exporting firms’ transition rates and the proportion of output destined for the export market suggests that Zimbabwean exporters face no serious problems in entering export markets, though the analysis of the direction of exports shows that entry barriers are a problem for firms that want to export to Europe or the United States. The most acute problem in raising Zimbabwe’s exports lies not in the inability to access foreign markets but in increasing both the number of firms which are exporters and increasing the output shares sold on the export market. Exporters seem to shun specializing into exporting while many firms who should be able to operate in foreign markets (large firms especially) fail to do so. Half of the country’s manufactured exports find their way into the regional market, a third go the European market and the remaining proportion is absorbed by USA and the rest of the world. Hence less than half of the exported output is sold in the markets with the highest purchasing power. This together with the substantial proportion of large firms that are not exporters suggests that a large potential is left untouched. The results present a case for a broadly based export strategy, designed to improve international competitiveness on the basis of multiple (deliberately created) comparative advantages. They indicate that the relevant question is not one of accepting or rejecting market-based reforms, but rather one of assigning importance to institutional factors which facilitate access to these markets. A more stable political and economic environment is potentially very effective in inducing export oriented investments and helping firms to operate less hesitantly on the export market. Additional institutional support for trade promotion to help entrepreneurs develop their network will also improve export performance. The challenge is therefore not one of making a choice between the market and government or between cooperation 25 and competition but to get the balance right between economic incentives and institutional support. 26 References Benson, C. and E. Clay (1994), The impact of drought on Sub-Saharan economies, IDS Bulletin, vol 25, No. 4, 24-32. Biggs. T, M. Miller, C. Otto and G. Tyler (1996), Africa can Compete! Export Opportunities and Challenges for Garments and Home Products in the European Market, Washington, D.C. World Bank. Bigsten. A, P. Collier, S. Dercon, B. Gauthier, J.W. Gunning, J. Habarurema, A. Isaksson, A. Oduro, R. Oostendorp, C. Pattillo, M. Soderbom, F. Teal and A. Zeufack (1998), Exports of African Manufactures: Macro Policy and Firm Behavior, Journal of International Trade and Development (forthcoming). Botchwey, K., P. Collier, J.W. Gunning and K. Hamada (1998), Report of the group of independent persons appointed to conduct an evaluation of certain aspects of the enhanced structural adjustment facility, IMF. Brainard, S.L. (1993) A simple theory of multinational corporations and trade with a trade-off between proximity and concentration, NBER working paper series No. 4269. Calvo, G.A. (1988) Costly Trade Liberalizations, Durable Goods and Capital Mobility, IMF Staff Papers, Vol 35 no.1: 461-473. Clerides, S.K., S. Lach and J.R. Tybout (1998), Is learning by exporting important? Micro-dynamic evidence from Colombia, Mexico, and Morocco, Quarterly Journal of Economics 113(3): 903-947. Corell, B. and I. Welch (1996), Culture, information and screening discrimination, Journal of Political Economy, 104(3): 554-271. DeBresson C. and F. Amesse (1991), Networks of innovators: A review and introduction to the issue, Research Policy, vol 20: 363-379. Dixit, A. and V. Norman (1980) Theory of International Trade, Cambridge, Cambridge University Press. 27 Dixit, A. (1989), Entry and exit decisions under uncertainty. Journal of Political Economy 97(3):620-638. Fafchamps, M. (1998), Market emergence, trust and reputation, Stanford University, mimeo. Fafchamps, M. (1998a), Ethnicity and credit in African manufacturing, Stanford University mimeo. Government of Zimbabwe (1992), Study of Monopolies and Competition Policy in Zimbabwe, Report Submitted to the Inter-Ministerial Committee on the Monopolies Commission. Grabher, (1993) The embedded firm: On the socioeconomics of industrial networks, London, Routledge Greene, W. (1993): Econometric Analysis, Macmillan Publishing Company, New York. Gunning, J.W. (1996) Trade Policy Review of Zimbabwe, in S. Arndt and C. Milner (eds.) The World Economy: Global Trade Policy 1996, Blackwell Publishers. Harrold. P., M. Jayawickrama and D. Bhattasali (1996), Practical Lessons for Africa from East Asia in Industrial and Trade Policies, World Bank Discussion Paper No. 310. Heckman, J.J. Detecting discrimination (1998), Journal of Economic Perspectives 12(2): 101-116 Hirschman, A.O. (1958), The Strategy of Economic Development, New Haven. IMF, Balance of Payment Statistics, various issues. Krugman, P. (1992), Does new trade theories require a new trade policy?, The World Economy, vol. 15(4): 423-441. 28 Lou Egan, M. and A. Mody (1990) Buyer-Seller links for export development, Industry and Energy Department, Working Paper No. 23, The World Bank. Mafusire, A. (1998) Growth, exports and foreign direct investment: The Zimbabwean experience, Paper presented at the Conference on: Zimbabwe: Macroeconomic Policy, Management and Performance since Independence(1980-1998): Lessons for the 21st Century, Harare. Mlambo, K. and M.Z. Ncube (1998), Real and Monetary Determinants of Real Exchange Rate Behavior in Zimbabwe, Paper Presented at the Conference on: Zimbabwe: Macroeconomic Policy, Management and Performance Since Independence (1980-1998): Lessons for the 21st Century, Harare. Nadvi, K. (1998), Knowing me, knowing you: Social networks in a surgical instrument cluster of Sialkot Pakistan, IDS discussion paper no. 364. Ncube, M., P.Collier, J.W. Gunning and K. Mlambo (1999), Trade Liberalization and Regional Integration in Zimbabwe, in A. Oyejide, B.Ndulu and J.W. Gunning (eds), Regional Integration and Trade Liberalization in Sub-Saharan Africa Volume II, Basingstoke and London: Macmillan. Nurkse, R. (1953), Problems of capital formation in underdeveloped countries, London. Ohlin, B. (1933), Interregional and International trade, Cambridge, Cambridge University Press. Pindyck, R.S., (1988). Irreversible investment, capacity choice and the value of the firm, American Economic Review 78 :969-985. Ratnayake, R. (1994), Effective protection of manufacturing industries in Zimbabwe, A report prepared for the Monitoring and Implementation Unit of the Economic Structural Adjustment programme. Roberts M.J and J.R. Tybout (1997), The Decision to Export in Colombia: an Empirical Model of Entry with Sunk Costs, American Economic Review 87: 545-563. 29 UNCTAD (1995), Trade, Environment and Development, Lessons from empirical studies, the case of Zimbabwe, Trade and Development Board, ad hoc working group on Trade, Environment and Development, mimeo. Vernon, R. (1966), International lnvestment and lnternational Trade in the Product Cycle, Quarterly Journal of Economics, vol 80(2), 190-207. World Bank (1995), Trends in Developing Countries. 1 At certain levels networks may become counterproductive. Especially if firms solely rely on network contacts, better informed businesses who make use of both network and market information may become more profitable. 2 lndirect exporters are considered to play an important role in export promotion. For example input- supplying indirect exporters are said to be critical for establishing backward linkages from exports, whilst output-supplying indirect exporters are critical for developing firms, for example trading companies, that specialize in overseas marketing and identifying cheaper sources of inputs supplies, technology, etc. Experience from the East Asian exporting countries, particularly South Korea, shows that the proper identification of indirect exporters, and the provision of inputs at prices to both direct and indirect exporters were the key factors that partly explain their export success story (Harrold et al. 1996). 3 The first list of imports which were put on the OGIL list in October 1990 were tinplate, plastic raw materials, dyes and related chemicals for the textile industry and klinker for cement production. However, the initial unanticipated side effects of OGIL was the massive increase in imports of goods found on the list. Some of the imports were non-essentials, such as large quantities of plastic toys from the Far East that found their way onto the Zimbabwean domestic market. 4 Items which were initially listed on the OGIL negative list included such non-essential consumption goods like household electronic goods, clothing items, beverages, precious and semi-precious stones, arms and ammunition, etc. 5 In the Zimbabwean case, for the years after 1992 no updates on the dollar index could be obtained. For this use is made of the real effective exchange rates. If the real exchange rate depreciates by 20% between 1992 and 1995, the dollar Index for 1992 is multiplied by 0.8 to yield a relative price estimate for 1995 (Bigsten et al. 1998). 6 Available statistics (see Benson et al 1994) show that due to the 1991/2 drought, manufacturing output declined by 9.3% in 1992. There was also a 6% reduction in the foreign currency receipts from manufactured exports and a 2% reduction in export receipts. 30 7 On 1st May 1992, South Africa increased its duty rates from 20% to 35% for yarn, from 20% to 50% for fabrics and from 30% to 100% for clothing. Until then South Africa was Zimbabwe’s single largest textiles and clothing market. Later in November 1993 these rates were scaled down slightly to 32%, 45% and 90% for yarn, fabrics and clothing respectively. 8 Note however that the figures in table 9.1 reflect calendar years while for figure 9.2 book years are used. 9 Ostrich exports provide a telling example of what entry costs entail. By 1994 ostrich meat was in high demand in Europe and the Zimbabwean ostrich industry was ready to enter this market. To be able to do so, first a European approved abattoir had to be built. Unfortunately there was considerable uncertainty concerning the specifications if only because there was no European Union approved ostrich abattoir anywhere in the world. Not only are the abattoir requirements unclear the rules of ostrich trade are also largely still open as the EU had still not decided whether ostriches were poultry or wildlife (UNCTAD, 1995). 10 Measured as the average export intensity at the firm level. 11 The reason is that networks are assumed to reduce the fixed costs of international transactions. Agents with networks are more likely to be picked by potential customers and thus have to invest less in their international sales operations. The size of the transactions which are concluded is however independent of the incorporation in a network. 12 In the furniture industry for example firm product ranges typically include bedroom, dining sets and office furniture made in small quantities -as few as 10 to 20 pieces- often using 40 year old machinery. A similar situation holds in the textile industry where investments in specialized equipment for different types of clothing makes no economic sense, unless a firm starts to specialize in export production. Then investments in special appliqués for ironing presses to increase the speed of pressing shirt sleeves or collars and more sophisticated parts for sewing machines such as customized sewing foots are needed to make construction more exact (Biggs et al 1996). With these types of equipment, export goods can be produced by specialising in bulk production, so that export goods can be sold at low profit margins. 13 The introduction of capacity utilisation on the right hand side of the equation could lead to an endogeneity problem if higher exports lead to increased capacity utilisation. In principle we could control for this endogeneity by using some kind of instrument or introducing lags. However, it is safe to assume that the endogeneity problem is small, given the low export intensity of the surveyed firms. 14 The implementation of financial sector reform before the fiscal deficit was under control led to a period of very high real interest rates and a very large increase in the government interest bill (Botchwey et al. 1998). 31