Report No. 42296-TZ Zanzibar The Effect of the Investment Climate on Performance of Micro and Small Enterprise in Zanzibar A Comparison with Mainland Tanzania and other Countries October 2007 Document of the World Bank ACKNOWLEDGEMENTS .............................................................................................. Table of Contents 1 EXECUTIVESUMMARY ............................................................................................... 2 INTRODUCTION .............................................................................................................. 4 I THEWORLDBANKENTERPRISE I1.. SURVEY .................................................. 4 THEINVESTMENT CLIMATEINTANZANIA CHAPTER2: CHALLENGES THAT SMALLISLAND ECONOMIESFACE ........................................................69 CHAPTER3: FIRMPERFORMANCEINZANZIBAR ............................................ 18 I LABOR .. PRODUCTIVITY .............................................................................. 18 I1 CAPITAL PRODUCTIVITY ........................................................................... 22 I11 CAPACITY UTILIZATION ........................................................................... 23 I V.. TOTALFACTOR PRODUCTIVITY ............................................................... 24 CHAPTER4: THE INVESTMENT CLIMATE INZANZIBAR ............................... 27 I FIRMSPERCEPTIONSABOUTTHEINVESTMENTCLIMATEINZANZIBAR 11 .. ..27 MACROECONOMIC INSTABILITY ............................................................... 28 111 TAXRATESANDADMINISTRATION 29 I V .. .......................................................... CORRUPTION AND REGULATION ............................................................... 30 V FINANCE ..................................................................................................... 32 V I.. TRAINING AND WORKER SKILLS .............................................................. 33 CONCLUSIONS .............................................................................................................. 36 STATISTICAL APPENDIX ........................................................................................... 39 REFERENCES ................................................................................................................. 42 ANNEX1: SELECTEDEXAMPLESOF SEZDEVELOPMENT ................................... 45 ANNEX2: QUESTIONNAIRE ................................................................................... 48 ACKNOWLEDGEMENTS This report was written by George Clarke, Tilahun Temesgen, and Michael Wong. It i s based on an analysis o f World Bank Enterprise Survey (WBES) data conducted by the Economic and Social Research Foundation (ESRF) inDar es Salaam, Tanzania and the Regional Program on Enterprise Development (WED) at the World Bank incollaboration with the National Bureau o f Statistics (NBS). The surveys were conducted between April and July 2003 and inJuly 2004. The ESRF staff members involved in this project included Professor H.K. Amani, Josaphat Kweka, Oswald Mashindano and John Kajiba. John Paton played an important role inthe first round o f survey implementation. Johannes Hoogeveen and Paolo B. Zacchia acted as peer reviewers. Vijaya Ramachandranprovided helpful comments. 1 EXECUTIVESUMMARY Zanzibar i s a small island economy close to the Tanzanian mainland. As with other small economies, Zanzibar i s vulnerable to terms o f trade and other shocks. Although the economy i s slowly becoming less dependent on clove production and clove products, this natural resource still makes up a large share o f exports andjobs. Diversifying into manufacturing would allow Zanzibar to reduce this vulnerability. This goal is consistent with Zanzibar's Growth Strategy (2006-2015), which sees manufacturing as one of four economic growth sectors. This study looks at factors that affect the performance o f manufacturing enterprises-and their resulting incentives to invest. The main source o f information is a 2003-2004 World Bank Enterprise Survey. The report i s complementary to an earlier report that looked at the investment climate in the whole o f the United Republic o f Tanzania, including Zanzibar, using data from 2003-the 2004 Investment Climate Assessment (Regional Program on Enterprise Development, 2004). To avoid redundancy, this report focuses primarily on areas where the investment climate i s different inZanzibar than inmainland Tanzania. The firms covered in the survey are poorly integrated into international markets as few export outside Tanzania. Because Zanzibar's economy is small, firms that remain focused on local markets likely will stay small and employ fewer than 50 workers. Even compared to small firms in other small economies, firms in Zanzibar appear to be especially unlikely to export. For example, they are less likely to export than are similar firms in other small island economies such as the Seychelles or Mauritius. This suggests that the poor export performance i s not simply due to small country size. Other factors also play a role. One important factor is that productivity is low, meaning firms are not highly competitive on international markets.Labor productivity i s very low in Zanzibar, even when compared to labor productivity in mainland Tanzania. Whereas the median firm in mainland Tanzania produces over $2000 o f value-added per worker, the median firm in Zanzibar produces only $1000 per worker. Moreover, when compared to the best performing countries in Sub-Saharan Africa labor productivity in mainland Tanzania i s low. Although low labor productivity in Zanzibar can, in part, be explainedby the fact that firms are smaller than their counterparts on the mainland and labor productivity i s typically lower in small firms, micro and small enterprises (MSEs) are less productive than MSEs on the mainland and inmost other countries inSub-Saharan Africa. Another reason for the low productivity i s that MSEs tend to be less capital intensive in Zanzibar. The median MSE in Zanzibar has less than $300 o f capital per worker, compared to about $1400 on the mainland. This figure i s also low when compared to other countries in Sub- SaharanAfrica. For example, in Senegal and Kenya the median MSE i s over $3000 o f capital per worker. After controlling for their smaller size and lower capital intensity, firms in Zanzibar compare more favorably with firms elsewhere in Tanzania. But total factor productivity-the part o f 2 productivity that cannot be explained by the use o f capital and labor-remains lower for the average enterprise in Zanzibar than for the average enterprise in mainland Tanzania. Further, total factor productivity i s far lower inZanzibar than inmost productive low-income countries in Sub-Saharan Africa. Most aspects o f the investment climate appear similar in Zanzibar to mainland Tanzania. But there are some differences. Zanzibar often compares favorably with mainland-including in areas such as corruption and tax administration that mainland firms see as serious growth impediments. For example, Zanzibar firms were less likely to need to bribe officials in order to secure government contracts or to `get things done'; tax administration i s less burdensome; and the regulatory burden is lighter. Although this gives some reason for optimism, problems remain. First, wages and salaries are lower in Zanzibar than on the mainland. Whereas the medianMSE in Zanzibar reports that it pays about $400 per worker per year, the medianenterprise in Dar es Salaam reports that it pays $950 per worker and the medianMSEelsewhere inmainland Tanzania reports that it pays $1700 per year. Although low wages allow MSEs inZanzibar to remain competitive despite low labor productivity, this i s at the expense o f workers. Iffirms could improve their productivity, wages could increase thus allowing firms to remain competitive. One plausible reason for low wages and low labor productivity i s the workforce's low skill base. Worker skills was one o f the few areas o f the investment climate that firm managers inZanzibar found to be a greater obstacle to their enterprise's operations and growth than their counterparts on the mainland. Objective data confirm these perceptions; both workers and managers have less education in Zanzibar than elsewhere in Tanzania. Overall, firms in mainland Tanzania compare relatively unfavorably with firms in Kenya and Uganda with respect to employee, but not manager, education. Despite this, enterprises inZanzibar invest less inworker training than enterprises inmainland Tanzania do. Another area o f the investment climate that appears to be a larger problem inZanzibar is access to finance. Managers perceive access to credit as a greater constraint, firms are less likely to have loans or overdraft facilities, and firms without loans are less likely to say that they did not want one inZanzibar than inmainland Tanzania. When combined with the previous finding that firms in Zanzibar have less capital per worker, it suggests that access to finance might be even more problematic inZanzibar than it i s elsewhere inTanzania. In summary, Zanzibar's investment climate appears relatively favorable in some respects. Despite this, wages and labor productivity remain low and firms are not very competitive. Two areas that appear more problematic inZanzibar than elsewhere in Tanzania-and are also worse inTanzania than inthe better performing low-income countries inAfrica-are access to finance and worker skills. If the Government could improve the investment climate along these lines, wages could increase while allowing firms to remain competitive. 3 INTRODUCTION Zanzibar i s a small island economy in the Indian Ocean. It i s close to the Tanzanian mainland andconsists o f two main islands, Unguja and Pemba, and several smaller islands. Both its geographic area and population are small. About 1,072,000 people lived in Zanzibar in 2005 andthe combined area o f the islands is only about 2,654 square kilometers. Zanzibar is a semi- autonomous part o f the UnitedRepublic o f Tanzania. It has its own Government, which consists o f a legislature, a House o f Representatives; an executive, headed by the President o f Zanzibar; and a Judiciary. Over the past decade, Zanzibar's growth has been fairly high, averaging about 7 percent a year. But it also has been unstable. For example, GDP growth exceeded 16 percent in 1996, but fell to about 2 percent in 1998 (Ministry o f Finance and Economic Affairs, 2006). Population growth has also been rapid, averaging about 3 percent a year. Agriculture remains the most important economic sector, contributing between 21 and25 percent o f GDP over the past decade (Ministry o f Finance and Economic Affairs, 2006). It also contributes about 40 percent o f jobs and 70 percent o f foreign exchange earnings. Manufacturing i s less important, contributing only about 5 to 6 percent o f GDP and less than 5 percent o f export earnings. Zanzibar's Growth Strategy (2006-2015), however, recognizes that manufacturing could be an important source o f growth (Ministry o f Finance andEconomic Affairs, 2006). This study looks at factors that constrain investment and growth in manufacturing. Since an Investment Climate Assessment (Regional Program on Enterprise Development, 2004) was completed recently for the entire country o f Tanzania, which included Zanzibar, this report mostly focuses on areas where Zanzibar is different from the mainland. The results from the Tanzania Investment Climate Assessment are summarized below. I. THEWORLDBANKENTERPRISESURVEY The main source o f information for this report is data from firm surveys conducted in 2003 and 2004 inmainland Tanzania and Zanzibar. The 2003 survey, which covered about 276 manufacturing firms, was conducted between April and June 2003 and included 10 regions in mainland Tanzania and Zanzibar. The 2004 survey, which covered only Zanzibar, was conducted in July 2004 and included 19 manufacturing firms. Firms in the 2003 survey were omitted from the 2004 sample frame. The surveys were conducted by the Economic and Social Research Foundation (ESRF) in Dar es Salaam and the Regional Program on Enterprise Development (WED) at the World Bank, incollaboration with the National Bureau o f Statistics W S ) . Firmsinthe two surveys were randomly selected from a sample frame that was stratified by firm size and location. The sample frame was constructed using lists o f firms with over 5 employees from various government sources, including a list from the National Bureau o f Statistics. To ensure that the sample included enough large firms, large firms are overrepresented (i.e,, the probability o f selection depends on firm size). The regions and sectors 4 covered in the survey were selected based on the concentration o f manufacturing firms in these areas. There were a total o f 40 manufacturing enterprises from Zanzibar inthe two surveys-21 inthe 2003 survey and 19inthe 2004. Because ofthe modest size ofthe two Zanzibar surveys, we pool the two samples for most o f the analysis. In cases where there are significant differences betweenthe 2003 and 2004 surveys, we note this inthe text. Although the sample i s small, it represents a significant share o f manufacturing firms on the island. The 24 firms with over 10 employees account for about 20 percent o f firms this size. In contrast, the survey includes only nine firms with between five and nine workers (about 3 percent o f firms inthis size class). The remaining firms had fewer than five employees-about 0.3 percent o f firms are inthis size class. These firms employed about 15 percent o f workers in manufacturing inZanzibar at the time o f the surveys.' At the time o f our survey about 50 percent o f firms inthe Zanzibar ES were inthe food and beverage sector, with another one-third in the wood and furniture sector (see Table 1). These sectors are also important in the survey in mainland Tanzania, although they make up smaller shares o f our sample. More firms from Zanzibar are inthe construction materials sector, with fewer represented bytextiles or the garment industries. Survey firms from Zanzibar are smaller than firms from mainland Tanzania (see Table 1). Almost one-third o f the sample from Zanzibar has fewer than 10 employees, compared to 14 percent o f firms from Dar es Salaam and 10 percent o f firms from elsewhere in mainland Tanzania. There are no large firms inthe sample for Zanzibar. Incomparison, large firms make up over aquarter o fthe sample from themainland. Because the firms from Zanzibar are smaller on average than the firms from the mainland, some comparisons are difficult. Most notably, labor productivity and capital intensity (that i s the capital the firm has for each worker) tend to be lower inmicro and small enterprises (MSEs). Moreover, MSEs often face different investment climate constraints than do larger firms. substandard infrastructure. Because o f these concerns -- andto make the Zanzibar and mainland They often find it more difficult to get financing and find it harder to cope with Tanzania samples comparable -- our analysis concentrates only on MSEs. 5 Table 1: Characteristics of enterprisesinthe Enterprise Surveys for Zanzibar and mainland Tanzania Zanzibar Dar es Other Salaam Mainland Numberof Enterprises 40 112 143 Micro (1-9 workers) 34% 14% 10% Small (10-49 workers) 54% 37% 49% Medium(50-99workers) 11% 22% 14% Large (100-499 workers) 0% 20% 23% Very Large (500 andup) 0% 7% 4% Firmexports goodoutsideTanzania 3% 29% 30% FirmimportsrawmaterialsfromoutsideTanzania 11% 69% 48% Majority foreign-owned 3% 10% 4% Majority government-owned 0% 8% 4% FoodandBeverages 43% 22% 32% FurnitureandWood 35% 21% 23% ConstructionMaterials 15% 5% 3% Textiles andGarments 5% 10% 13% Other 3% 42% 29% 11. THEINVESTMENTCLIMATE INTANZANIA This report uses data from two surveys-one that covered mainland Tanzania and Zanzibar and one that covered Zanzibar only. Results from the first survey that covered the mainland and Zanzibar were presented in an earlier assessment (Regional Program on Enterprise Development, 2004). The earlier assessment did not present or discuss results for Zanzibar separately. To avoid redundancy, this report focuses on those areas o f the investment climate where there are significant difference between Zanzibar and mainland Tanzania. To put the results in context, this section summarizes the main results from the earlier assessment. As in that assessment, results in this section for Tanzania pool firms from mainland Tanzania and Zanzibar. Labor productivity i s low in Tanzania (including Zanzibar). In2002, value added per worker was about $2300 (in constant 2005 US$). This figure was considerably lower than in Kenya (about $5000 for each worker), but higher than in Uganda (about $1600).2 Comparisons with more recent surveys elsewhere in Sub-Saharan Africa confirm that labor productivity is lower in Tanzania than inmany other countries in Sub-Saharan Africa. Value-added per worker is lower in Tanzania than it i s in 20 o f the 32 countries in Sub-Saharan Africa where Enterprise Surveys had been done by December 2006. Even ignoring the six middle-income countries (South Africa, Namibia, Mauritius, Botswana, Swaziland and Cape Verde), this puts Tanzania somewhere close to the median for low-income countries in Sub-Saharan Africa and far below the fast growing economies o f China and India. Consistent with the idea that Tanzanian firms are not highly competitive, they were also less likely to export than firms inChina and Kenya-although more likely to export than firms in 6 Uganda. The difference between Tanzania and Kenya does not appear to be entirely explained by differences in firm size or sector o f operations. Even after controlling these, firms in Tanzania were less likely to export. Trade is also restricted by two other things-low competitiveness and the high burden o f trade and customs regulation^.^ Exporters rated trade regulations as the sixth greatest problem they faced. Moreover, consistent with this, customs and port delays were higherinTanzania (7 days for exports) than inKenya, Uganda or China. One o f the reasons for firms' low productivity and competitiveness in Tanzania is that worker skills and education are low. Workers in Tanzania have less education than do workers ineither Kenya or Uganda. About 43 percent hadonly aprimaryeducationor less at the time of these surveys in contrast to about 20 percent in Kenya and Uganda. Since the percentage o f workers with university-level education was about the same in all three countries, the gap is caused by the low percentage o f Tanzanianworkers with secondary andvocational education. Despite these dispiriting statistics, firms do not invest much in improving workers' skills. Although enterprises in Tanzania were more likely to have formal training programs than in Uganda, they were less likely to have them than either Kenya or China. Enterprises without training programs were most likely to say that they were unable to afford formal training programs or that informal training was enough. Contrary to this belief, formal training appears to payoff in Tanzania. Total factor productivity (TFP) was 11percent higher in the Tanzanian enterprises with formal training programs. As well as being asked for quantitative information to estimate levels o fproductivity and competitiveness, firms were asked what they see as the biggest problems they face. Enterprises inTanzania, including Zanzibar, were most likely to rate tax rates, electricity, interest rates, tax administration, corruption, access to finance, and macroeconomic instability as the most serious problems. As previously noted, exporters were also concerned about trade and customs regulations. Although perceptions provide an interesting starting point for the analysis, it is more difficult to make cross-country comparisons based on perception-based data than with more quantitative evidences4 For this reason, the survey also collected quantitative evidence on many areas o f the investment climate. This quantitative evidence, and evidence from other sources, suggests there is room for improvement inmany areas. For example, although Tanzania performs well on most measures o f governance (such as political stability, the rule o f law, and regulatory quality), corruption appears to remain a problem. About 33 percent o f enterprises that did business with the government said bribes are needed to secure government contracts and about 35 percent o f enterprises said that bribes also are needed to move things forward in customs, taxes, licenses, andother government services. Similarly, objective measures o fpower and financial sector performance also suggest that these are problems in Tanzania. Although the cost o f power i s not excessive compared to other countries in the region, reliability is a serious problem. The median enterprise in Tanzania reported losing 5 percent o f production because o f surges and outages in 2003. In comparison, 7 median firms in China and Uganda reported losing 0 percent of sales; firms in Kenya reported losing 5 percent of sales. Access to credit also appears to beworse than inKenyaor China. 8 CHAPTER 2: CHALLENGES THAT SMALL ISLAND ECONOMIES FACE Over the past 40 years, many economists and development experts have claimed that small economies-and especially small island economies-face more serious development challenges than larger economies. A recent study that looked at economic growth and development in small states, including small island economies, found many examples o f papers, conferences and seminars devoted to the problems these countries face (Easterly and Kraay, 2000). The authors noted that many featured titles included words such as `problems', `vulnerabilities', and `challenges'. These papers suggest several reasons why small island economies might not perform as well economically as larger non-island economies. Many o f these revolve around two issues- size and rem~teness.~Although some o fthe problems that they cause are macroeconomic, others might directly affect enterprise behavior, performance or structure. This section will discuss the ways inwhich size and remoteness affect economic performance and describe how these factors might affect enterprises inZanzibar. It will also present evidence from the Zanzibar Enterprise Survey on these issues. One way that size might affect economic performance i s that small economies might be more vulnerable to economic shocks. Economies o f scale and little diversity innatural resources prevent small economies from diversifying production across industries and sectors (Commonwealth Secretariat and World Bank Joint Task Force on Small States, 2000; Streeten, 1993), This makes them vulnerable to terms o f trade shocks, and natural disasters that affect the entire economy, or demand shocks (for example, shocks that affect global tourism). Easterly and Kraay (2000) show that small economies are more vulnerable to such shocks than larger economies. Growth volatility and volatility o f terms o f trade are significantly higher for small economies.6 Zanzibar could be vulnerable in this respect. Cloves and clove products dominate the economy, earning about 70 percent o f export earnings and employing more than 60 percent o f the labor force (Zanzibar Investment Promotion Agency, 2004). Although this could be a problem, Zanzibar is not the only economy in Sub-Saharan Africa that faces this vulnerability. Many countries inthe region export only a few primaryproducts (Collier, 1998). A recent study noted that inthe late 199Os,39 o f 47 o f African countries depended on two primary commodities for over half o f their export earnings (Morrissey and Filatotchev, 2000). As a result, most countries in the region-including larger economies-are susceptible to terms-of-trade shocks. Keeping this in mind, however, concentration does seem to be highin Zanzibar when compared to the Sub-Saharan Africa region. Diversifyinginto manufacturing, which makes up only about 5 percent o f exports, would lessen Zanzibar's vulnerability inthis respect. A second problem is that small size might restrict competition, especially in island economies where imports are more expensive. Because scale economies mean only a few domestic firms can operate in some sectors, it i s possible that production will be highly concentrated in at least some economic sectors. This could lead to higher prices, lower quality, or less innovation. Because firms from mainland Tanzania, including Dar es Salaam, have 9 access to Zanzibar's market, this might be a lesser concern in Zanzibar. Close to half o f firms from Dar es Salaam sell some o ftheir output inZanzibar suggestingthat this mightbe the case. Inpractice, competition does not appear to be especially low inZanzibar (see Figure 1). Finns from Zanzibar were more likely to say they had more than five competitors and that they faced competition from both domestic and foreign competitors than were firms on the mainland. They also report lower average local market shares than elsewhere in Tanzania (28 percent in Zanzibar, compared to 32 and 34 percent inDar es Salaam and elsewhere inmainland Tanzania). Moreover, the differences between Zanzibar and elsewhere in Tanzania are not statistically significant. This provides little support for the ideathat competition i s low inZanzibar. Figure 1 Comnetition i s no lower inZanzibar than inmainland Tanzania. 100% 75% W v) v) w- 2 50% s0 25% 0% 1 Morethan 5 competitors Any foreign competitors Any domestic competitors Source: World Bank Enterprise Surveys Note: Data is for MSEs only. Another problem small economies face i s that they might be unable to reach economies o f scale in production o f public and private goods. For public goods, many services might be subject to indivisibilities that make them costly to provide in small countries (Alesina and Spoloare, 1997). Consistent with this, the median wage bill for the public sector was about 10 percent higher in small developing countries than inlarger developing counties (Commonwealth Secretariat and World Bank Joint Task Force on Small States, 2000). Because public administration is more expensive, this can lower the quality or quantity o f public goods small countries pr~vide.~ For example, some studies suggest it i s more costly to provide education and training in small countries and that it might not even be financially feasible to provide some specialized or higher level education (Commonwealth Secretariat and World Bank Joint Task Force on Small States, 2000). Although it seems reasonable that economies o f scale might raise the cost o f providing public services such as health care or education, there i s little empirical data to support the idea that these services are worse in small economies. In fact, some studies find that health and 10 education outcomes appear to be better, not worse, in small economies. On average, child mortality rates are lower, life expectancy higher and school enrollment rates larger in small economies than in similar large economies (Dommen, 1980; Easterly and Kraay, 2000). One possible explanation i s that voters may be more willing to provide local goods, such as public health and education, in small economies ifthey are less diverse than larger economies (Alesina and others, 1999). Zanzibar also appears to perform well in this respect. Literacy rates are slightly higher and infant and child mortality are slightly lower in Zanzibar than in mainland Tanzania (Revolutionary Government o f Zanzibar, 2006) Difficulties reaching scale economies might also affect private production. With only small domestic markets to trade in, firms might have to produce less than they would ideally unless they are able to enter export markets. Further, it will also be difficult for countries to develop clusters o f firms in single industries-especially without makingthe country vulnerable to external shocks. For example, Streeten (1993) notes that "[flew small countries can afford an aircraft industry, integrated motor car production, or the production o f heavy railway stock." To the extent that this is the case, it suggests small economies must generally integrate themselves into international supply chains inat least some industries. Although the World Bank Enterprise Surveys for Tanzania and Zanzibar over-sampled large firms, the companies surveyed in Zanzibar were small compared to mainland firms. Most had fewer than 50 employees (see Table 1). The difference in samples reflects differences inthe overall population o f firms on the mainland and inZanzibar. Although many enterprises inboth Zanzibar and mainland Tanzania are microenterprises with fewer than 10 employees, there are fewer medium-sized or large firms with over 50 employees inZanzibar. Inthe Industrial Census o f 2001, only 2 enterprises in Zanzibar had more than 100 employees (about 2 percent o f enterprises with more than 10 employees) and none had more than 200 employees. In comparison, in2003-5, there were 88 firms inDar es Salaam with more than 100 employees (20 percent o f firms with more than 10 employees) and 17 with more than 500 employees (4 percent).' Because firms in Zanzibar are small, they might behave differently than firms on the mainland. First, small firms are less likely to export than larger firms.g The large fixed costs associated with setting up an international distribution or service network make exporting easier for large enterprises. Further, large enterprises have better access to finance than small enterprises-especially in developing countries-making it easier for them to finance these costs. Evidence from the World Bank Enterprise Survey suggests that firms from Zanzibar are less well integrated with the global economy than firms from mainland Tanzania (see Figure 2). Only one firm in Zanzibar reported having either direct or indirect exports outside Tanzania. Interestingly, however, this does not seem to be only because firms from Zanzibar are smaller. None o f the MSEs from Zanzibar exported outside Tanzania. In comparison, 12 percent o f MSEsfrom mainlandTanzania exported some part oftheir production. 11 Figure2 Firms inZanzibar are less well integrated with the global economy than firms from mainland Tanzania. 100% 75% .-E UJ LL 50% 'c 0 a? 25% 0% Firm exportsgood outside Firm imports raw materials Majorityforeign-owned of Tanzania from outside of Tanzania Dar es Salaam Other Mainland I Source: World Bank Enterprise Surveys Note: Data are for MSEs only. One possible reason firms from Zanzibar do not export, even after controlling for size, i s that Zanzibar does not border other Sub-Saharan Africa countries. Since African port facilities and procedures are often slow and overloaded, it is difficult to export overseas (Clarke, 2005). Perhaps because o f this, most manufacturing firms in mainland countries that export do so to neighboring countries. For example, manufacturing firms in the World Bank Enterprise Survey for mainland Tanzania were more likely to export to Kenya and Malawi than they were to any other countries. Since Zanzibar does not have land borders, firms in Zanzibar might find it especially difficult to export. If this were the case, we might expect to see similar patterns in other small island economies, especially those in Sub-Saharan Africa. Although many studies have shown that small economies, including small island economies, are open to trade, exports are often concentrated in a few industries (Commonwealth Secretariat and World Bank Joint Task Force on Small States, 2000). To see if Zanzibar looks like other small island economies, the percent o f firms exporting and exports as percent o f sales are shown for firms in Zanzibar and firms in several other small island economies where World Bank Enterprise Surveys were done (see Figure 3). Although firms in some island economies, such as Cape Verde, do not export much, firms in other countries, such as Mauritius, show success in this respect (see box 1). This suggests that Zanzibar does perform worse in export capacity than do other small island economies. 12 Figure 3: FirmsinZanzibar are less likely to export than firms inother islandeconomies. All firms MSEs only Mauritius Mauritius Maldives Maldives Dominican Dominican Republic Republic Cape Verde C a p Verde Zanzibar Zanzibar 0% 20% 40% 60% 80% 0% 20% 40% 60% % of firms exporting % of firms exporting Source: World Bank EnterpriseSurveys Note: Data varies between2002-2005, dependingon survey periodfor eachcountry. A final possibility is that the low share o fexports mightbebecause firms only `export' to mainland Tanzania. As definedinthe World Bank Enterprise Survey, exports only refer to those outside Tanzania. Using this methodology, sales Zanzibar firms make on the mainland are classified as domestic sales. Although the survey questionnaire did not include questions that allow us to identify sales in mainland Tanzania as distinct from sales in Zanzibar, results from other surveys suggest this fact might explain, in part, differences between Zanzibar and other small island economies. In a survey conducted in 2006, about 18 percent o f firms in Zanzibar export to the mainland. This would make Zanzibar less o f an outlier compared to other island economies with respect to exporting-although it would still perform less well than the best performing economies such as Mauritius. 13 Box 1: Mauritius' Success inExporting Mauritius once depended on a few agricultural exports for most o f its foreign exchange earnings. Even though Mauritius is a small, remote island, Mauritius now has a large manufacturing sector, especially in the textiles and clothing sectors. Manufacturing sector growth in Mauritius is often associated with its export-processing zone (EPZ) established in the early 1970s using Taiwan, China as a model. The fiscal incentives to investors, such as tax holidays or duty drawbacks, are common to other countries where EPZs exist. But unlike other EPZs, over 50 percent of investment came from local entrepreneurs. Rather than creating economic enclaves with plants set up by foreign multinational corporations with little local spillover effects, the EPZ in Mauritius has driven local manufacturing sector growth with forward andbackward linkages. The EPZ started showing positive results in the early 1980s. In 1971, there were only nine EPZ firms. By 2000, there were over 500. Employment in the EPZ also grew, from 644 workers in 1971to over 90,000 in2000. This gave sugar factory owners an alternative activity. This, in turn, carried them through the agricultural crisis in the late 1970s. It also built confidence and raised wages in other sectors o f the economy and brought foreign investors, global business linkages, andnew ideas into the Mauritius. These factors allowed Mauritians to modernize and to buildinvestor confidence inmany sectors, especially tourism. Source: Regional Programon Enterprise Development (2007) Zanzibar performs better after taking exports to mainland Tanzania into account. But firms from Zanzibar are still less likely to export to mainland Tanzania than firms from mainland Tanzania are to Zanzibar. About 44 and 39 percent of firms from Dar es Salaam and elsewhere on mainland Tanzania sell some o f their output inZanzibar. This is considerably higher than the 18percent of firms from Zanzibar that sell goods onthe mainland. So what can Zanzibar do to encourage manufacturing exports? Some things related to improved firm productivity and competitiveness are discussed in detail in the next Chapter. Other things, such as capitalizing on informal networks that Zanzibari's have in other regions, for example inthe MiddleEast. Trade and customs regulations, which are a serious problem in Tanzania overall, could also be improved (Regional Program on Enterprise Development, 2004). Finally, the Mauritian experience might suggest that Zanzibar's exports could be increased by settingup new Export Processing Zones (EPZs), or expand existing ones. Although EPZshave sometimes beensuccessful, such as Mauritius and Madagascar, they have not always performed well in Africa. Most zones are performing below expectations and some have failed (for example, see the description o f Senegal in Annex 1). Historically the main reasons for failure were inadequate provision o f infrastructure services, lack o f a market strategy, a public rather than a public- private partnership approach, and an inadequate institutional framework in these zones. This strongly suggests that merely setting up the zones is not enough. A second major issue that affects the economic performance o f many small island economies i s remoteness. One way remoteness might affect economic performance and firm behavior i s the effect distance has on transportation costs. Since small island economies are far from major sea and air routes, and all imported goods must be shipped or flown in, it can be 14 costly to import and export goods to small islandeconomies. Problems associated with physical remoteness are made worse by small economic size. Because small countries often need small shipments, bulk cargo shipments must be broken into smaller parts. Small, remote islands also are vulnerable to market exploitation by freight carriers and airline companies so as to further pushup transport costs. Consistent with the view that transportation costs are a problem, recent studies found that the ratio o f insurance and freight costs to merchandise imports i s highin small island economies (Atkins and others, 2000; Briguglio, 1995). The high cost o f transportation will raise the price o f imported goods, reduce the purchasing power o f consumers and raise market power for domestic firms inthese economies. Zanzibar i s less remote than many other islands (for example, manyPacific islands, Cape Verde or Mauritius). It i s found about 35 miles from the coast o f the mainland and i s only about 45 miles from Dar es Salaam. Although, as noted earlier, the poor performance o f ports might make importing and exporting more difficult, especially in Sub-Saharan Africa, it i s not as remote as some other small islands. In this respect, remoteness is less o f an issue in Zanzibar thanfor other islandeconomies. Although the World Bank Enterprise Survey does not provide direct information on transportation costs, it does provide some indirect information. Firms are asked about delays in getting shipments o f inputs, damage during shipping, and about number o f days o f inventory o f important inputs. Briguglio (1995) argues that firms inremote locations keep larger inventories to avoid delays ifthey runout. The indirect measures o f transportation costs do not suggest that transportation problems are larger in Zanzibar than in mainland Tanzania. Firms from Zanzibar report slightly higher average losses because o f transportation delays than firms on the mainland (4.3 percent o f sales compared to 3.6 percent on the mainland), but the difference is not statistically significant. They also report lower average losses because o f breakage, theft, and spoilage during transportation (0.7 percent o f the value o f cargo compared to 1.6 percent on the mainland), andreport keeping, lower inventories o f needed inputs (median o f 7 days compared to median o f 30 days). Only the second difference is statistically significant. Size and remoteness interact in important ways. Because the small size o f the domestic economy makes it difficult to find domestic producers, firms are often more dependent on imported inputs than firms in larger countries. That is, even though the high cost o f transportation raises the price o f inputs, firms often need imported inputs because o f costly or unavailable domestic substitutes. The high cost o f imported inputs will, in turn, affect firm performance by raising costs and, inso doing, make it more difficult for them to export. Despite high transportation costs, firms in small island economies are highly dependent on imported inputs. For example, firms from Cape Verde, Mauritius and the Seychelles are all heavily dependent on imported inputs-as are firms in small economies on the mainland such as Swaziland, Botswana and Namibia+ompared to larger economies on the mainland (see Figure 4)* Unfortunately, the Zanzibar Enterprise Survey does not provide similar information. Although the survey asks about use o f imported inputs from outside Tanzania, it does not 15 separate inputs from mainland Tanzania from inputs from Zanzibar. As a result, it i s unclear whether domestic goods are from the mainland. Other evidence, however, i s consistent with this idea that imported inputs are important. For example, imports are about 80 percent o f its basic requirements, including goods brought in from mainland Tanzania, Zanzibar (Zanzibar Investment Promotion Agency, 2004). Figure4: Firmsinsmall economies, includingsmall islandeconomies, are heavilydependentupon importedinputs 8o 1 v) al 60 c0 8? v) m 40 2g 20 .-E 0 Source: World Bank EnterpriseSurveys. Note: All values are means for enterpriseswith availabledata. Data varies between 2002-2005, dependingon survey period for each country. So what is the net impact o f size and remoteness on macroeconomic performance? Although most theoretical studies suggest that small economies face more serious challenges thanlarger economies, recent empiricalwork has challenged this pessimistic view. Easterly and Kraay (2000) find that small economies have higher per capita GDP, have better health and education outcomes, and are more productive than larger economies. They also find that after controlling for other factors, small economies grow faster. Earlier studies found weaker results, i.e., that small economies grew at about the same pace as larger economies. Armstrong and others (1998) find neither small states nor islands grow more slowly than large and non-island economies. Milner and West (1993) find the same. The mixed empirical results can be explained in various ways. Most notably, several authors noted the benefits o f being small. Easterly and Kraay (2000) find that small countries benefit more from being more open to trade thanthey lose because o fmacroeconomic instability. Armstrong and others (1998) suggest other explanations, including that social cohesion might be enhanced andthat flexibility inpolitical decision-makingmightbe higher. In summary, although Zanzibar's small size appears to affect firm characteristics and behavior in some ways (for example, size and internationalization), the effect does not appear 16 large inother ways (most notably, transport costs and competition). Moreover, it is not clear that Zanzibar's size will affect its economic performance. Although the literature on small island economies has focused largely on problems and vulnerabilities o f small islands, there i s little empirical evidence that small islands perform worse ineconomic terms than other countries. 17 CHAPTER3: FIRMPERFORMANCEINZANZIBAR The previous chapter noted that firms in Zanzibar were smaller and less well integrated into international markets than firms in mainland Tanzania. A natural question therefore i s whether this affects firm performance. This chapter addresses this issue, comparing firm performance in Zanzibar with firm performance in mainland Tanzania, Kenya and Uganda, and other low-income economies in Sub-Saharan Africa. I. LABOR PRODUCTIVITY Labor productivity, a basic measure o f firm productivity, is the output a firm produces less the cost o f raw materials (such as iron or wood) and intermediate inputs (such as engine parts or textiles) divided by the number o f workers used to produce the output. Labor productivity i s higher in firms that produce more output with fewer raw materials and fewer workers. Differences in labor productivity can be the result o f differences in technology, organizational structure, worker skills, management style and ability, or in differences in the capital available to a firm. Because the labor productivity measurement does not take capital (Le., machinery and equipment) into account, it will be higher infirms that use capital inplace o f labor (Le,, firms that are capital intensive). Figure 5: Labor productivity is lower on average inZanzibar that it is inother parts of Tanzania and lower than elsewherein Sub-SaharanAfrica $9,000 s3 5 I $6,000 L a a D UQ) U 9 lll $3,000 -Q) >m 3 $0 a2 s Source: World BankEnterprise Surveys Note: All values are mediansfor enterpriseswith availabledata. Value addedis calculatedby subtractingintermediateinputs and energy costs from salesfrom manufacfuring. Workers includebothpermanentandtemporary workers. Data were collectedbetween 2002 and 2005 dependingon survey periodfor each country. Data collectedprior to 2005 are convertedto 2005 figuresusingGDP deflators. Values are convertedto US$ usingaverage exchangeratesfor 2005 from WorldDevelopmentIndicators(World Bank, 2007b). 18 Labor productivity (value-added per worker) is lower for the median firm in Zanzibar than for firms elsewhere in mainland Tanzania (see Figure 1). Firms in Zanzibar produce less thanUS$lOOO (in2005 US$) of value-added per worker, compared to $2900 per worker inDar es Salaam and $2400 per worker elsewhere inmainland Tanzania. This i s also lower than value added per worker in other low-income Sub-Saharan African countries. Although labor productivity is only slightly higher in a few countries (Ethiopia, Gambia, Guinea-Conakry), in manycountries it is over two times as high; inthe most productive low income countries such as Kenya and Senegal it can be more than four times as highas inZanzibar. One reason for Zanzibar's low labor productivity is that firms in Zanzibar tend to be smaller than firms inmainland Tanzania and elsewhere inAfrica. Among the countries inFigure 1, labor productivity is lower among MSEs than it is for larger enterprises in nearly all the countries for which data are available. After controlling for this by only looking at micro and small enterprises (MSEs) with less than20 employees, the difference betweenZanzibar and elsewhere inmainland Tanzania is less stark. Median labor productivity i s about $1000 per worker inZanzibar, $1930 inDar es Salaam and $2300 elsewhere inmainland Tanzania. Labor productivity for MSEs is higher in Zanzibar than in Ethiopia, similar to Guinea-Conakry, the Gambia, Madagascar and Uganda, but it remains significantly lower than in better performing African countries such as Kenya and Senega1. Figure 6: Wages are low inZanzibar Source: World BankEnterpriseSurveys Note: All values are mediansfor enterpriseswith availabledata. Labor cost is the total cost of wages and salariesand allowances, bonusesandother benefitsfor bothproductionand administrativestaff. All valuesare for MSEsonly. Employees include permanent and temporary workers. Datawere collectedbetween2002 and2005 dependingon survey periodfor each country. Data collectedprior to 2005 are convertedto 2005 figures usingGDP deflators. Values are convertedto US$usingaverageexchangerates for 2005 from WorldDevelopmentIndicators(World Bank, 2007b). 19 Given that labor productivity i s so low in Zanzibar-even after controlling for firm size-a natural question is how firms manage to stay in business despite their low productivity. An important factor is that wages are also low-about 40 percent lower for MSEs in Zanzibar than they are for MSEsinDar es Salaam. Wages also are low compared to other low-income countries in Sub-Saharan Africa. This is consistent with qualitative evidence from a small survey o f foreign investors in Zanzibar that found that wages were not an inhibiting factor for investment and that they were low compared to mainland Tanzania (Office o f Chief Government Statistician, 2005). Although labor costs in mainland Tanzania are broadly comparable with other low-income countries in Sub-Sahara Africa, labor costs are lower in Zanzibar than in most other countries (see Figure 6), especially the most productive African countries. The strong cross-country correlation, even among low- income countries, between wages and firm productivity confirm the link between labor productivity and wages (0.75). Why are labor productivity and wages and salaries lower in Zanzibar than they are elsewhere in Tanzania and in other African countries? One reason might be that workers' educational attainment i s low-even compared to mainland Tanzania. Zanzibar's Growth Strategy notes that inthe entire workforce (Le., not only among workers inMSEs), close to two- thirds o fworkers are categorized as unskilledworkers andthat few workers have universitylevel education (Ministry o fFinance and Economic Affairs, 2006). Educational attainment also seems to be low for the enterprises in the World Bank Enterprise Survey. About 18 percent o f MSE employees in Dar es Salaam and 11 percent of MSE employees elsewhere in Tanzania have a tertiary education, compared to only 0.2 percent have the same inZanzibar. This is also lower than inKenya or Uganda-where about 12percent o fMSE employees have a tertiary education . Figure 7: Educationalattainment is lower inZanzibar than on the mainland. 100% 75% 50% 25% 0% Zanzibar Dares Other Kenya Uganda (MSEs) Salaam Mainland (MSEs) (MSEs) I Primaryor Lower Secondaryor Vocational 17University I Source:InvestmentClimate Assessments 20 The difference between Zanzibar and mainland Tanzania is much smaller for primary education-about 54 percent o f workers inMSEs in Zanzibar have a primary education or less, compared to 56 percent inother parts o fmainland Tanzania and 47 percent inDar es Salaam. As noted earlier, far fewer workers inKenya or Uganda -where most workers have some secondary or vocational training-, have solely a primary education. This suggests the main difference between Zanzibar and mainland Tanzania i s intertiary education-fewer workers progress from secondary to tertiary levels in Zanzibar. But the main difference with Kenya and Uganda i s at the level o fprimaryratherthansecondary education. What is the net impact o f this on firm competitiveness? Because both wages and productivity are low in Zanzibar, likely as a result o f low education, firms could potentially remain competitive despite having lower labor productivity than firms in other low income economies. Unit labor cost (labor cost as a percent of value-added) is a measure of labor costs that measures the net impact o f labor costs on competitiveness by taking into account differences in productivity. Unit labor cost i s higher when higher labor costs are not hlly reflected in higher productivity. When a firm's unit labor cost is high (Le., when labor costs are high compared to productivity), it will find it more difficult to compete on international markets. Although unit labor cost i s not the only factor that affects competitiveness-for example, it does not take into account the cost o f capital or capital intensity-it i s a better measure o f competitiveness than labor costs alone. Figure 8: Althoughwages are low, productivity is even lower meaningunit labor costs are high. gu 50% m 3 n 40% 0 30% v 3 20% 5 10% m 0% I I . . , . Source: World BankEnterpriseSurveys Note: All values are mediansfor enterpriseswith availabledata andfor MSEsonly. Datawere collectedbetween2002 and 2005 dependingon survey periodfor each country. Data collectedprior to 2005 are convertedto 2005 figures usingGDP deflators. Values convertedto US%usingaverage exchangeratesfor 2005 from WorldDevelopment Indicators (World Bank, 2007b). 21 Inthiscase, lowwages do not appearto makeupfor lowproductivity. Unitlaborcosts, which are equal to 48 percent, are higher for MSEs inZanzibar than for similar firms inDar es Salaam or mainland Tanzania (see Figure 8). It i s also highcompared to most other countries in Sub-Saharan Africa, where unit labor costs are usually between 20 and 30 percent. Although lower than inthe countries with the highest unit labor costs, such as Uganda and Guinea-Bissau, this suggeststhat firms inZanzibar will find it difficult to compete on internationalmarkets. The extent to which this i s the case will depend inpart on how capital intensive firms inZanzibar are. 11. CAPITAL PRODUCTIVITY The low educational attainment o f workers in MSEs in Zanzibar may partially explain why labor productivity is low inZanzibar. But it is not the only reason. Inaddition to worker education, labor productivity is affected also by the capital a firm has. Labor productivity is usually higher for capital intensive firms and industries. This fact suggests it is important to also look at capital use. Although measures o fcapital intensity provide some context for the previous results, it i s more difficult to measure capital than it is to measure labor (for example, it is easy to measure wages andnumber o fworkers). This is because most machinery is long-lived, providing services over a long period. As a result, it is difficult to measure its contribution to output in one year. As capital ages, it becomes less productive (i.e., it depreciates in value) even before it becomes obsolete or stops working. Although accounting rules for depreciating machinery and equipment exist, these often bear little resemblance to true rates o f economic depreciation-and can vary across countries. The book value o f capital (Le., the value o f capital in company accounts) i s therefore not an especially accurate measure o f the value o f capital-especially for small firms that often do not keep detailed audited accounts. As an alternate measure o f the value o f capital, recent World Bank Enterprise Surveys have asked firm managers how much it would cost to replace their equipment in its current condition. Although this i s a u s e h l measure o f capital-and provides another check on results- inpractice, markets for used capital are thin. Firmmanagers might not know the true value of their capital-especially ifthe equipment is oldor ifthey havenot purchased similar equipment in years. This is the measure of capital that this assessment focuses on-although results are qualitatively similar when looking at book value. Low capital use might also explain the low productivity o f firms in Zanzibar. Capital intensity, capital per worker, i s lower for the median MSE in Zanzibar (about US $300 per worker) than inDar es Salaam ($2500) or elsewhere inmainland Tanzania (about $1500). This i s also low compared to other countries in Africa-although not the lowest. For example, firms in Uganda have about $700 of capital per worker and firms in Senegal and Kenya have significantly more capital (over $3500per worker). 22 Figure9: Micro and SmallEnterprisesinZanzibar haveless capitalperworker than similar enterprisesinmainlandTanzania andelsewhereinAfrica $6,000 62 v, 3 Lo 0 $4,000 Source: Investment Climate Assessments Note: All values are medians for all MSEswith available data. Capital is the net book value of machinery and equipment at the end of 2002. Workers include both permanent and temporary workers. See earlier figures for explanation of exchange rates and deflators. Although capital per worker gives an idea about how capital intensive firms are, it does not provide much information on how productively that capital is being used. Capital productivity, the ratio o f value added to the net book value o f machinery and equipment, measures how productively firms use capital. It i s analogous for capital to (the inverse of) unit labor costs for workers. Capital productivity i s higher in firms that produce a lot with little machinery and equipment. This could be because the firm is more efficient or it could be because the firm uses a labor intensive production technology (i.e., relies heavily on labor to produce their output). In contrast to labor productivity, which measures the value-added per worker, capital productivity (value added over capital) is highinZanzibar. The ratio o f value added to capital i s about 180 percent inZanzibar, significantly higher than inDar es Salaam (90 percent) and other parts of Tanzania (100 percent). It is also significantly higher than in Kenya or Senegal-two countries that use capital intensively-but slightly lower than in Uganda, another country with modest capital per worker. Given that the median firm in Zanzibar has little capital, it i s not surprising that capital productivity is high. 111. CAPACITYUTILIZATION Another reason why labor productivity might be low in Zanzibar is that firms may have much unused capacity. This, however, does not seem to be the case. As part o f the World Bank 23 Enterprise Survey, enterprise managers were asked to estimate capacity utilization in their firm- how large their actual production was compared to the maximum amount that they could have beenproducedwith the capital and workers that they employed at the time. The average MSE in Zanzibar reported that its capacity utilization was about 60 percent (see Figure 10). This is slightly higher than elsewhere in Tanzania-about 55 percent and 50 percent for Dar es Salaam and other mainland locations. But it is still lower than inmore successful low-income economies in Africa, such as Senegal (70 percent) and other successful economies such as China (70 percent). Figure10: CapacityUtilizationis higher inZanzibar than elsewhereinAfricaand inmost middle incomecountrie-but lagsbehindthe mostproductiveregionsinChina. loo 1 Zanzibar Dares Other Uganda Kenya Senegal Salaam Tanzania Source: World BankEnterprise Surveys. Note: All values are mediansfor enterpriseswith available data. Capacity utilizationis directly reportedby enterprise managersand is defined as the amount of output actually producedrelative to the maximum amountthat could beproducedgivencurrent capital stock and employment. All values are for micro and small enterprises only. IV. TOTAL FACTOR PRODUCTIVITY Although the results presented in the previous subsection provide useful measures o f performance and competitiveness, they have some drawbacks. The main problem i s that when considered separately, labor and capital productivity can present incomplete, and sometimes contradictory, evidence. For example, in Zanzibar labor productivity is low, while capital productivity is high. Both o f these are due, at least in part to the fact that firms are labor intensive. Total factor productivity (TFP) avoids some o f these problems by taking capital and labor use into account simultaneously. Differences in total factor productivity are the result o f differences in things other than capital or labor. For example, differences might be because of differences in firm organization, differences inmanagement efficiency, or differences inworker skills or education. To the extent that differences in technology are not embedded in the 24 machinery and equipment that the firm uses, differences in total factor productivity can also account for this. Besides taking capital and labor use into account, TFP has several other advantages over the other measures o f firm performance presented in the previous section. Most importantly, TFP is calculated in a regression framework, therefore it is possible to control for many things duringcalculations. For example, when comparing average TFP across countries it is possible to control for differences insector composition. Some problems, however, remain. 1. As with labor productivity, labor costs per worker, capital per worker and other measures o f firm performance denominated in U S dollars, cross-country comparisons o f TFP are vulnerable to exchange rate fluctuations. Ifthe exchange rate is overvalued compared to its long-run equilibrium then TFP might look artificially low. Measures that are ratios, such as capital productivity or unit labor costs, avoid this problem. 2. As discussed earlier, capital i s more difficult to measure than labor for theoretical and practical reasons. Since TFP depends on measurement o f capital, it will be mismeasured when capital is mismeasured. 3. Because estimates are calculated in a regression framework, it is less clear than the measures in the previous subsections. One issue is that estimates o f TFP for groups o f firms do not have natural units. For cross country comparisons, TFP i s shown as % o f TFP in South Africa-one o f the most productive countries inAfrica. A second issue is that estimates can depend up estimation method (for example, ordinary least squares, frontier estimation, or least absolute deviations (LAD) estimation). Inpractice, however, the results in this section do not appear to be highly sensitive to different estimation techniques. Total factor productivity i s low inZanzibar, suggestingthat the low labor productivity is not simply the result o f low capital intensity, size or sector. This suggests that other factors are playing a role (for example, low capacity utilization or low education and skills among workers). Notably, total factor productivity is significantly lower thaninmainland Tanzania. 25 Figure 11: Total Factor productivityis also low inZanzibar 40% $ .E 30% E c sO2 20% v 10% 0% Source: Investment Climate Assessments Note: Calculated as a residual from a least absolute deviations (LAD) regression. 26 CHAPTER 4: THEINVESTMENTCLIMATE INZANZIBAR Labor and total factor productivity are lower in Zanzibar than elsewhere inTanzania and than in most other low-income countries in Sub-Saharan Africa. Further, firms in Zanzibar are far less productive than firms in the most productive low-income economies in Africa (e.g., Kenya and Senegal) or in the fastest growing low-income countries such as China. Moreover, wages remain low in Zanzibar even in comparison with elsewhere in Tanzania." Iffirms were able to improve their productivity, they would be able to increase wages and salaries while increasing competitiveness ininternationalmarkets. Recent work shows how steps to improve the investment climate can result in increased firm productivity and improved firm performance.'' Firms in Africa appear to be especially disadvantaged inthis respect; investment climate problems mean that indirect costs are far higher for firms in Africa than for firms in other countries and their productivity is consequently lower.l2 As noted earlier, this chapter focuses on areas o f the investment climate inZanzibar that are different from mainland Tanzania. The results for Tanzania as a whole are discussed briefly inChapter 2 andinmore detail inRegionalProgramonEnterprise Development (2004) I. FIRMSPERCEPTIONSABOUTTHEINVESTMENTCLIMATEINZANZIBAR So what are the most significant investment climate problems in Zanzibar? The World Bank Enterprise Surveys ask enterprise managers what they rate as the greatest constraints on enterprise development and growth. Although perception-based measures have several well- known problems, they provide a useful starting place for analyses o f the investment ~1imate.l~ Throughout this report, the perception-based data are supplemented with objective indicators o f the investment climate to ensure robustness. Figure 12 shows the percent o f firms that rated various components o f the investment climate as a major or very severe obstacle to enterprise operations and growth. Several things standout clearly inthe data. Most notably: 0 Perceptions about obstacles are similar in Zanzibar and elsewhere in Tanzania. The top five obstacles inboth Zanzibar and mainland Tanzania were tax rates, tax administration, cost o f financing, access to financing, and electricity. 0 Perceptions in Zanzibar were generally more favorable than in mainland Tanzania- enterprises in Zanzibar were less likely to rate most areas o f the investment climate as serious obstacles than enterprises inmainlandTanzania. 0 But there were some differences. Enterprises inZanzibar were far less concerned about tax rates and administration, corruption, and macroeconomic instability than firms in mainland Tanzania. In most of these areas, the difference in perceptions remains statistically significant after controlling for size (see Appendix). 0 Inone area, access to finance, perceptions were far less favorable in Zanzibar. The difference i s statistically ~ignificant.'~ 27 0 Consistent with the earlier evidence that workers inZanzibar are less well educated than workers in mainland Tanzania, firm managers in Zanzibar remain more worried about worker skills than managers elsewhere in Tanzania. The difference is not statistically significant. Figure 12: Inmostcases, firms inZanzibar and mainland Tanzania had similar perceptions about obstacles to enterprise operations and growth -a 0 .n m 0 .zm8 3 75% m $j 50% m m g! m gF 25% a g! 0% S m Note: Figure shows the percent of MSEs that rated a particular area as a 'major' or 'very severe' obstacle to enterprise operations and growth 11. MACROECONOMICILITY I N S T A B The first area where there was a noticeable difference between Zanzibar and mainland Tanzania is macroeconomic instability-enterprises in Zanzibar are far less concerned about it than enterprises inmainland Tanzania. This might seem puzzling since the two main aspects of macroeconomic instability that are mentioned on the survey, exchange rate instability and inflation, are similar in both locations. As a result, we might expect perceptions also to be similar. But differences exist between the two samples. Most notably, whereas firms from mainland Tanzania were interviewed in 2003, half o f the firms from Zanzibar were interviewed in 2004. This appears to partly explain the difference. Among the MSEs interviewed in Zanzibar in 2003, about 30 percent said that macroeconomic instability was a major or very 28 severe obstacle. Incontrast, only about 12percent o f the MSEs interviewed inZanzibar in 2004 said the same. This does not, however, explain the entire difference. About 40 percent o f MSEs in mainland Tanzania rated macroeconomic instability as a serious obstacle in 2003- higher than the number inZanzibar inthe same year. In addition to the level of inflation or exchange rate volatility, other factors affect whether firms see macroeconomic instability as a serious problem. For example, firms involved ininternationaltrade were more concerned about macroeconomic instabilitythanwere firms that are not. Whereas 49 percent o f firms that directly imported raw materials and 43 percent o f firms that export ratedmacroeconomic instability as a serious obstacle, only 30 and40 percent o f firms that did not directly import or export rated it as the same. Since MSEs in Zanzibar were less likely to import and export goods from outside o f mainland Tanzania and Zanzibar (3 percent and 0 percent) than MSEs inmainland Tanzania (44 percent and 12 percent), this might also account for part o f the difference. To see whether this i s the case, we estimated an econometric model to study whether observable differences between firms inmainland Tanzania and Zanzibar explain the difference inperceptions. After controlling for size, sector of operations, age and whether the enterprise imports or exports, the difference between Tanzania and Zanzibar becomes small and statistically insignificant, suggesting that the differences in perception are due to observable differences between firms (see Appendix). 111. TAXRATESAND ADMINISTRATION Taxation i s another area that MSEs from Zanzibar were less likely to rate as a serious obstacle. Although the government o f Tanzania has tried to improve tax administration inrecent years, this i s not why firms inZanzibar ratedtax administration as a lesser obstacle than firms on the mainland did.'5 Firms inZanzibar were no less likely to rate tax administration as a serious problem in2004 than they were in2003-about 26 percent rated it as a serious obstacle in2004, compared to 29 percent in 2003. Moreover, the difference in perceptions is not statistically significant for firms in 2003 and 2004 (Le., the difference could be random due to the small sample).16 Fewer firms rated tax rates as a serious problem in 2004 than in2 0 0 3 4 7 percent in 2004 compared to 67 percent in 2003. But once again, after controlling for other factors that affect perceptions about tax rates (e.g., enterprise size and sector o f operations), the difference i s not statistically significant. Moreover, in both cases, enterprises on the mainland were more likely to rate tax administration as a serious problem even after controlling for survey year and enterprise characteristics. Objective measures o f the investment climate support the idea that tax administration i s a less pressing problem on Zanzibar than it is in mainland Tanzania (see Figure 13). MSEs in Zanzibar reported they had fewer tax inspections and required meetings than firms in either Dar es Salaam or other mainland locations. Whereas the medianMSE inZanzibar reported only two required meetings or inspections by tax officials inthe previous year, the median MSE inDar es Salaam re orted seven meetings and the median MSE elsewhere on the mainland reported 6 meetings." Firms also were less likely to report when bribes or gifts were requested or needed 29 duringthese meetings (about 5 percent inZanzibar compared to 19percent inDar es Salaam and 27 percent inother mainlandlocations).'* Despite having fewer requiredmeetings and inspections, tax evasion does not appear to be a more significant problem inZanzibar than elsewhere inTanzania. As part of the Enterprise Survey, managers were asked to estimate how much o f their revenues firms like theirs would report to the tax authorities. The question was asked inthis way so that managers could respond without incriminating themselves. MSEs in Zanzibar estimated that firms like theirs would report about 71 percent o f revenues to the tax authorities, compared to 68 percent in Dar es Salaam and 67 percent inother parts o f mainland Tanzania. This suggests that the increasing the numbero finspections andmeetings does not automatically improve compliance. Figure 13: Tax administrationis less burdensomeinZanzibar than elsewhere inTanzania. Zanzibar Zanzibar I l l 0 2 4 6 8 0% 10% 20% 30% IWMediannumberof tax inspectionsfor MSEsI Source: World Bank Enterprise Survey The higher levels o f tax evasion are consistent with the observation that bribes to tax officials are more common in mainland Tanzania. To the extent that firms use bribes to avoid paying taxes, it might not be surprising that both bribes and evasion appear more common in mainland Tanzania than inZanzibar Iv. CORRUPTION AND REGULATION In addition to taxation and macroeconomic instability, MSEs in Zanzibar were also far less likely to rate corruption as a serious obstacle than enterprises in mainland Tanzania. Whereas 53 percent o f managers o f MSE inmainland Tanzania said that corruption was a major or very severe obstacle to their enterprise's operations and growth, only 29 percent o f managers inZanzibar saidthe same. 30 Figure14: Bribes are less commoninZanzibar thaninmainlandTanzania. 1 % Dar 8s Salaam Zanzibar 2.1 0 0.1 0.2 0.3 0.4 0.5 0 2 4 6 HAverage percentof contractvalue needed in informal payments to secure government contract Source: WorldBankEnterprise Survey Note: Data is for MSEs only. As noted earlier, MSE managers in Zanzibar were far less likely to say that informal payments or gifts were requested during tax inspections than managers in mainland Tanzania were. Also consistent with this, they were far less likely to say that bribes or informal payments were `needed to get things done'. O f the 40 enterprise managers interviewed in Zanzibar, none said bribeswere needed to get things done-very different from Dar es Salaam and elsewhere in mainland Tanzania where 45 percent and 37 percent said bribes were needed to get things done (see Figure 14). MSEs also were less likely to report that bribes were needed to secure government contracts and, said whenbribes were needed, the payments were smaller on average. Why is corruption a less serious problem in Zanzibar than elsewhere in Tanzania? One possibility is that the regulatory burden i s lower inZanzibar. Whereas the median enterprise in Zanzibar had fewer than 10 requiredmeetings and inspections inthe year before the survey, the median enterprise in mainland Tanzania had over 15. Managers in Zanzibar also reported spending less time dealing with government regulations and inspections-about 5 percent o f senior management time inZanzibar compared with close to 15 percent inmainland Tanzania. Consistent with this evidence with respect to the regulatory burden, a recent study concluded that it takes less time and costs less to start a business in Zanzibar than in other areas o f Tanzania (World Bank, 2007a). Ittook only 129 days to complete all regulatory procedures to start a business inZanzibar, compared to 194 days inMbeya, 326 days inDodoma, and 528 days inKigoma. Themosttime-consumingsteps were transferringproperty, business registrationand business licensing. Other steps, such as connectingpower and water andNSSF registration were relatively less burdensome. 31 Over-regulation can lead to corr~ption.'~When the burden o f regulation i s high, managers have greater incentives to offer bribes to regulators and government officials to reduce the burden o f regulation. Moreover, government officials have greater reason to impose and enforce stringent regulations when they believe that they will be able to collect bribes from enterprise managers trying to avoid those regulations. To the extent that frequent meetings allow regulators and managers to develop close working relationships, the possibility o f collusion increases. V. FINANCE One area that enterprise managers in Zanzibar were more likely to see as a serious obstacle than their counterparts on the mainland was access to finance. Whereas 56 percent o f MSE managers in mainland Tanzania said that access to finance was a major or very severe problem, 68 percent o f MSE managers in Zanzibar said the same.20 Incontrast, there was little difference for cost o f financing-about 59 percent o f MSE managers in Zanzibar said it was a serious obstacle compared to about 58 percent o fMSE manager inmainland Tanzania. Given the high level o f concern about access to finance, it is not surprising that the financial sector i s not very developed in Zanzibar. According to the recent Zanzibar Growth Strategy (2006-2015), financial intermediation accounted for only 2 percent o f GDP in 2005 (MinistryofFinance andEconomicAffairs, 2006). The People's Bank o f Zanzibar (PBZ) dominates the banking sector inZanzibar. Of the 20 firms in World Bank Enterprise Survey that reported the primary bank that they did business with, 19 listed the People's Bank o f Zanzibar. The other banks operating in Zanzibar are branches o fmainbanks located inDar es Salaam. As o f late 2006, PBZ was operating under a Memorandum o f Understanding (MOU), which it entered into with the Government o f Zanzibar and the Bank o f Tanzania in2003. Under the MOU, PBZ was beingprepared for privatization (Ministryo f Finance and Economic Affairs, 2006). Objective data support the idea that financing i s an especially serious problem in Zanzibar-none o f the MSEs in the sample for Zanzibar reported that they had a bank loan, while 10 percent o f enterprises on the mainland outside o f Dar es Salaam and 19 percent o f enterprises in Dar es Salaam did (see Figure 15). Firms inZanzibar were also less likely to get credit from suppliers and were less likely to have overdraft facilities. Why do so few MSEs in Zanzibar have loans? O f the 30 MSEs in Zanzibar that responded to the question on whether they had a bank loan, 25 said that they had never applied for one. This was similar to in mainland Tanzania, where 89 o f 111 MSEs without loans said that they have never applied. Firmswere also askedwhy theyhadnever applied for a loan. They were allowed to give multiple responses to the question (that is, they could say that collateral requirements were too stringent and that interest rates were too high). The most common responses in Zanzibar were that collateral requirements were too stringent (50 percent o f MSEs), application procedures too cumbersome (56 percent) and interest rates were too high (48 percent). Incomparison, 52 32 percent o f MSEs in mainland Tanzania said that collateral requirements were too stringent, 57 percent said application procedures were too cumbersome, and 69 percent complained that interest rates were too high. Figure 15: MSEswere far lesslikelyto haveloansor overdraft facilitiesinZanzibar than in L mainland Tanzania. 70% 1--- _-_---- 60% 1 60% n I I 50% 43% 40% 30% 20% 10% 0% 0% 0% Other Mainland Percentof firms with overdraft facility 0Percentof firms with creditfrom suppliers Source: World Bank Enterprise Survey Note: Data is for MSEsonly. Cultural values that shunborrowing might partly account for the low level o f borrowing in Zanzibar, but this does not seem to drive the difference between mainland Tanzania and Zanzibar.21 Infact, MSEs inZanzibar were less likely to say that they had not applied for a loan because they did not want or need one than MSEs in mainland Tanzania.22 Whereas only 32 percent o f MSEs without loans inZanzibar said that they did not want one, 50 percent o f MSEs without loans inmainland Tanzania said the same. When combined with the previous information that MSEs in Zanzibar are less capital intensive than MSEs in mainland Tanzania, the evidence suggests that access to credit i s more difficult for MSEs in Zanzibar than it i s for MSEs in mainland Tanzania. Managers perceive access to credit as a greater constraint, firms are less capital intensive, firms are less likely to have loans or overdraft facilities, and firms without loans are less likely to say that they did not want a loan in Zanzibar than inmainland Tanzania. All this evidence suggests that finance i s a greater problem inZanzibar than elsewhere inTanzania. VI. TRAININGAND WORKER SKILLS Managers were slightly more likely to rate worker skills as a serious obstacle inZanzibar than they were in mainland Tanzania-although the difference was not statistically significant. Further, as noted in the previous section, objective indicators suggest that workers in MSEs in Zanzibar are less well educated thanworkers inMSEsinmainland Tanzania. 33 Managers also tend to have less education that their mainland counterparts. Only 6 percent o f MSE managers inZanzibar have a university education-far less than on the mainland where over 50 percent o f managers do (see Figure 16). MSE managers inZanzibar were slightly more likely to have vocational education than MSE managers on the mainland, but the difference was not large. Figure 16: Managers have lesseducation and less experience inZanzibar than managers on the mainland. 1 I I I Other Other Mainland Mainland Dares Dares Salaam Salaam 15.6 Zanzibar ~1 I Zanzibar 0 5 10 15 20 0 0.2 0.4 0.6 0.8 Average years of experience for manager in foreign enterprises IIAverageyearsofexperienceformanager ~ Source: World Bank Enterprise Survey Note: Data is for MSEs only. MSE managers in Zanzibar also had less experience than managers in MSEs on the mainland. The average manager had about 12 years o f experience working in the sector before runningthe establishment, with only 0.3 years inforeign-owned establishments. Incomparison, managers in Dar es Salaam had over 15 years experience with close to four years working in foreign-owned companies. The low level o f experience in foreign-owned companies is not surprising given that foreign-owned companies appear less common in Zanzibar than in mainland Tanzania. Finally, enterprises were also less likely to provide formal training to their employees in Zanzibar, Only 10 percent o f MSEs in Zanzibar reported a formal training program, compared to 26 percent in Dar es Salaam and 31 percent elsewhere on the mainland. When enterprises without formal programs were asked why this was, the most common responses were that training was not affordable or that in-house informal training was adequate for their needs. 34 Figure 17: MSEswere less likelyto provideformaltraining inZanzibar than inDar es Salaam or elsewhere in mainland Tanzania. I 40% I 31% 30% 20% 10% 0% Zanzibar Dar es Salaam Other Mainland I 0Percentof MSEswith formaltrainingprograms Source: WorldBankEnterpriseSurvey Note: Datais for MSEsonly. One reason why enterprises in Zanzibar might provide less training than enterprises on the mainland is that managers with less education tendto be less likelyto provide formal training to their workers. 31 percent o f enterprises in Tanzania with a manager with a university degree had a formal training program, compared to only 21 percent o f enterprises with managers without a degree. 35 CONCLUSIONS This study looks at firm performance and the investment climate inZanzibar. The main source o f information i s a 2003-2004 survey o f manufacturing enterprises. The report i s complementary to an earlier report looking at the investment climate inthe whole o f the United Republic o f Tanzania, including Zanzibar (Regional Program on Enterprise Development, 2004). To avoid redundancy with the earlier report, this report focuses on areas where the investment climate i s different inZanzibar from mainland Tanzania. Zanzibar is a small island economy found close to the Tanzanian mainland. Like other small island economies, limited diversification and a small domestic market make Zanzibar vulnerable to terms o f trade and other shocks. Diversifying into manufacturing-a goal that i s consistent with Zanzibar's Growth Strategy (2006-2015)-would reduce this vulnerability. This report looks at existing manufacturing firms inZanzibar to see how their performance compares with similar firms in other parts o f Tanzania, other countries in Africa, and other small island economies. It also compares the investment climate inZanzibar with the investment climate on the mainland andinother nearby countries. In many ways, the investment climate in Zanzibar appears relatively favorable when compared to the investment climate on the mainland. Firms are less likely to be concerned about most aspects o f the investment climate. Objective evidence i s consistent with the subjective evidence-the burden o f regulation appears lower in Zanzibar than in mainland Tanzania, tax administration i s less burdensome, and fewer firms report payingbribes. Despite this, firms in Zanzibar do not appear to be competitive. Few firms export any part o ftheir output-even to the mainland-suggesting that they cannot compete ininternational markets. Moreover, firms in Zanzibar are both small and unproductive when compared to firms from other countries in Sub-Saharan Africa, including mainland Tanzania. Although labor productivity i s low in Zanzibar partly because firms are small, MSEs are less productive even than similar firms on the mainland or elsewhere in Sub-Saharan Africa. Reasons for low productivity include the fact that MSEs do not use much capital and low worker skills and education. Low wages, coupled with even lower labor productivity, makes it difficult for Zanzibar firms to compete ininternational markets. Although the investment climate inZanzibar i s more favorable inmany areas than it i s on the mainland, firms remained more concerned about several areas. Consistent with the evidence on worker skills and capital intensity, firms were more likely to say that access to finance and workers skills and education were serious problems than firms on the mainland were. Objective data are consistent with this-fewer firms had loans, firms reported having less capital and workers and managers were less likely to be university educated in Zanzibar. Improving education-and taking steps to attract skilled workers from the mainland and elsewhere-and improving access to finance, therefore, should be a priority. Although poor access to finance might partly explain the low capital intensity in Zanzibar, other factors less easily captured in an World Bank Enterprise Survey might also play a role. One factor that might affect investment that is not captured easily in a firm survey is 36 political uncertainty and instability. Previous studies have shown that private investment i s lower in countries that are less politically stable (Stasavage, 2002). The observed instability in Zanzibar might therefore contribute to low investment. Inadditionto takingsteps to improve competitiveness, the government couldtake several additional steps to improve export performance. Although Tanzania has made progress with respect to improving trade and customs regulations in recent years-reducing the time to complete export procedures from 30 days in 2005 to 24 days in 2007 and the time to complete import procedures from 51 days to 30 days-these periods are lengthy compared to the best performing countries, where similar procedures often are completed in a week, or less. Singapore's example, in particular, shows that many procedures can be reduced significantly even in strong-performing economies. Other things, such as capitalizing on informal networks, for example inthe MiddleEast, that Zanzibari's have inother regions, would also be useful. Box: Reducingcustoms delays through computerizationinSingaporeand Ghana In recentyears, governmentshaveusedcomputerizationto dramatically reduceprocessing times for imports and exports. Rather than requiringsubmissionof multipleforms to multiple agencies, a trader now can electronicallysubmit a single document that containsall the informationrequired by different agencies. This information canthen be submittedto all relevant agencies, which then respondwith the necessarypermits or requestadditionalinformation. By eliminatingoverlappingrequirementsand multiple forms, the process reducestransactioncosts for firms and minimizesdirect contact betweenpublic officialsandthe trader, potentiallyreducingopportunities for side-payments. Singapore used these methodsin 1989to reduceprocessingtime from 2-4 days to a few minutesand the number of required documents frombetween3 and35 to a single document. Freightforwarders estimate that the programhas reducedtheir cost of handlingtrade documentationby between20 and35 percent. Singapore's success, and a similar programinMauritius, inspiredthe government of Ghanato adopt a similar programcalledTradeNet. Beforethe program, importersestimated that the fastest clearancetime at sea ports was four days, while the averageclearancetime was severalweeks. After implementingthe program, about 14 percentof clearancetook less than a day at Tema port and only 11percent more than five days. At the airport, averageclearancetimes fell from three days to four hours,with 18percentof clearances taking less than two hours. Although computerizationcan reducedelays, it will not succeed unlessproceduresare modifiedto fully exploit its potentialbenefits. BeforeimplementingTradeNet, the Ghanaiancustoms administration already was usinga standardsoftwarepackageto helpthem processimports. But procedureswere not designed to take advantage of the packageand as a result the technology was underused. For example, customs declarationshadto be manuallyenteredinto database, a processthat took up to 24 hours, rather than being submittedelectronically. Source: De Wulf (2004): World Bank (1998) More activist approaches to improving integration should be approached with caution. One possibility would be to start new, or expand existing, export processing zones. The rational for export processing and economic zones for most countries i s to boost competitiveness, reduce the cost o f doing business and to increase firm level competitiveness. Over 100 countries, from developed countries such as the USA or Canada to other developing countries such as Madagascar and Vietnam have applied the zones successfully. The traditional economic zone focuses on improving cost competitiveness by providing a package o f incentives such as import andexport duties exemptions andtax holidays. The zones are oftenrestrictedto isolatedenclaves 37 to relatively remote areas or near transport hubs. Today, most zones are shifting their target from low-margin, low cost-cost activities to higher value-added industries. With increasing globalization, zones are increasingly embedded in the existing supply structure and in the local economy. Zanzibar has established or plans to establish Free Economic Zones within Fumba Area, Amaan Industrial Park in Unguja and Michewweni Area in Pemba, in addition to Free Port in MaruhubiArea inUnguja. The zones provide exception from custom laws and regulations and other incentives provided with an investment certificate. What i s striking i s that the zone policy has so far had little impact on the Zanzibar export profiles. The majority o f traditional zone enterprises tend to be in labour intensive activities such as apparel, textiles and electronic assembly industries. Zanzibar so far has attracted trade merchandise (second hand clothes) for the mainland market. This is no different thanmany zones inislandeconomies inthe Caribbean. The Dominican Republic, Jamaica, and Barbados receive large foreign direct inflows from the USA andexport back to the same market. Inthe case o fZanzibar however, the purchasingpower o f the mainland does not fill the role o fthe leadingForeign Direct Investment (FDI) investor. Successful zones are characterized by some or all o f the following features: Streamlined regulatory framework Public-private partnership approaches for zone development Largely private sector-led; leadrole for one developer Clear zone designation and development criteria Top level, integrated support o f government e.g., Jordan, UAE Competition on the basis o f facilitation and services rather thanincentives Zone authority is autonomous, flexible, and focused on regulation, Regulatory authority capabilities are built up Minimization o f public expenditure by locating zones carefully/using existing facilities While experience to date inZanzibar is limited, it confirms the experience elsewhere that the success o f EPZs depends on the three main areas discussed above: (i)reliable infrastructure services at international comparable standards, (ii) an overall strategy in which the EPZs/SEZs are embedded, and (iii)adequate institutional framework that enables strong private sector participation andprofessional management. 38 STATISTICALAPPENDIX 1.1 Methodology. The methodology i s similar to the methodology used in a recent paper by Gelb, Ramachandran, Shah and Turner (2006). The question o f whether firms in Zanzibar have different perceptions about the investment climate is approached by estimating the following equation: Perception about IC, = p, +p2Zanzibar Dummy +p2Size+E, (3.1) The dependentvariables are dummyvariables indicating whether the manager o f firm irates that area o f the investment climate as a major or very severe obstacle. The independent variables are a dummy indicating whether the firm i s located in Zanzibar and firm size (number o f workers). Because the dependent variable is a dummy variable, the model is estimated using standard maximumlikelihood estimation. Results from the regression for each o fthe obstacles are shown below. Since the sample for Zanzibar is made up almost entirely o f MSEs, the regressions only include these firms. As an additional control for size differences, firm size (log o f number o f workers) i s included in the regression. This allows us to look at whether perceptions are differentinZanzibar and the mainland after controlling for firm size. In addition, we look more closely at differences in perceptions about macroeconomic instability by including extra regressions looking at this variable. Inparticular, these regressions control for whether the survey was conducted in2003 or 2004. 39 0 d REFERENCES Alesina, Alberto, RezaBaqir,andWilliamEasterly.1999. "Public GoodsandEthnicDivisions." Quarterly Journal of Economics 114(4):1243-1284. Alesina, Alberto, andEnrico Spoloare. 1997. "Onthe Number and Size of Nations." Quarterly Journal of Economics 112(4):1027-1056. Armstrong, H.,R.J. DeKervenoael,X Li,andR Read. 1998. "A Comparisonof the Economic Performanceof DifferentMicrostates,andBetweenMicro-StatesandLarger Countries." World Development 26(4):639-656. Atkins, J. P., S. A. Mazzin, and C. D.Easter. 2000. "CommonwealthVulnerabilityIndexfor DevelopingCountries." EconomicPaper40. CommonwealthSecretariat,LondonUK. Bertrand,Marianne,and SendhilMullainathan.2001. "DOPeopleMeanWhat They Say? Implicationsfor SubjectiveSurvey Data."American Economic Review: Papers and Proceedings 9l(2): 67-72. Bigsten,Arne, PaulCollier, StefanDercon,MarcelFafchamps,BernardGauthier, JanWillem Gunning,Abena Oduro, Remco Oostedorp, Catherine Pattillo,Mans Soderbom, Francis Teal, andAlbert Zeufack. 2004. "DOAfrican ManufacturingFirmsLearnFrom Exporting." Journal of Development Studies 40(3):115-141. Briguglio, Lino. 1995. "SmallIslandDevelopingStatesandTheir EconomicVulnerabilities." World Development 23(9):1615-1632. Clarke,GeorgeR. G. 2005. "BeyondTariffsand Quotas: Why Don'tAfricanManufacturers Export More?" PolicyResearchWorkingPaper3617. WorldBank, WashingtonDC. Collier,Paul. 1998. "Globalization: Implicationsfor Africa." InZubair IqbalandMoshinS Khan, eds., TradeReform and Regional Integration in Africa WashingtonDC: International MonetaryFund,pp. 147-181. Commonwealth Secretariat andWorldBankJoint Task Forceon SmallStates. 2000. "Small States:MeetingChallengesinthe GlobalEconomy." World Bank: WashingtonDC. Djankov, Simeon,RafaelLa Porta,FlorencioLopez-de-Silanes,andAndreiShleifer. 2002. "The RegulationofEntry." Quarterly Journal of Economics 117(1): 1-37. Dollar, David, Mary Hallward-Driemeier,andTayeMengistae.2003. "Investment Climate and FirmPerformance inDevelopingCountries." WorldBank: WashingtonD.C. Processed. Dommen,Edward. 1980. "SomeDistinguishingCharacteristics ofIslandStates." World Development 8(7-8): 931-943. Easterly,William, andAart Kraay. 2000. "SmallStates,SmallProblems?Income, Growth, and Volatility in Small States." World Development 28(11):2013-2027, 42 Eifert,Benn, Alan Gelb, andVijaya Ramachandran.2005. "BusinessEnvironmentand Comparative Advantage inAfrica: Evidencefromthe InvestmentClimate Data." Regional Programon EnterpriseDevelopment126. WorldBank, WashingtonDC. Escribano,Alvaro, andJ. Luis Guasch.2005. "Assessingthe Impactofthe InvestmentClimate on ProductivityUsingFirm-LevelData:Methodology andthe Cases of Guatemala,Honduras andNicaragua." World Bank: WashingtonDC.Processed. Gauthier, Jean-Paul. 2004. "Free Zones: Performance,LessonsLearnedandImplicationsfor Zone Development."ForeignInvestment Advisory Service: WashingtonDC. Processed. Gelb, Alan, Vijaya Ramachandran,ManjuKediaShah, and Ginger Turner.2006. "WhatMatters to AfricanFirms?The RelevanceofPerceptionsData." WorldBank: WashingtonDC. Processed. Grenier,Louise,Andrew McKay, and Oliver Morrissey. 1999. "Exporting, Ownership, and ConfidenceinTanzanianEnterprises." WorldEconomy 22(7):995-1011. Milner,C., and T. Westaway. 1993. "CountrySize andtheMedium-TermGrowthProcess: Some Country Size Evidence." WorldDevelopment21(2):203-212. MinistryofFinanceandEconomicAffairs. 2006. "Zanzibar'sGrowth Strategy (2006-2015)." The Revolutionary Governmentof Zanzibar:Zanzibar. Morrissey, Oliver, and Igor Filatotchev. 2000. Globalizationand Trade: The Implicationsfor ff ExportsFromMarginalizedEconomies." Journal of DevelopmentStudies37(2): 1-12. NationalBureauof Statistics.2006. "BusinessSurvey 2003-2005." NationalBureauof Statistics: Dar es Salaam,Tanzania. Office of Chief Government Statistician,Zanzibar. 2005. 2004 Zanzibar BusinessSurvey." Office M of Chief GovernmentStatistician:Zanzibar. --------,2007."2001 Censusof IndustrialProduction. Volume I:Analysisof Size 10+ Establishments." RevolutionaryGovernmentof Zanzibar: Zanzibar. Recanatini, Francesca, ScottWallsten, andLixinColinXu. 2000. "SurveyingSurveys and QuestioningQuestions: LearningfromWorldBankExperience." PolicyResearchWorking Paper 2307. WorldBank,Washington DC. RegionalProgramon EnterpriseDevelopment,AfricaPrivateSector Group. 2004. "Investment Climate Assessment:EnterprisePerformance andGrowthIn Tanzania." WorldBank: WashingtonDC. --------,2007.ffAnAssessmentof the Investment Climatein CapeVerde." WorldBank: Washington DC. Revolutionary Governmentof Zanzibar. 2006. "Zanzibar Strategy for Growth andthe Reduction of Poverty(ZSGRP)." RevolutionaryGovernmentof Zanzibar:Zanzibar. 43 Rhee, Yung Whee, andTherese Belot. 1990. "ExportCatalystsinLow-Income Countries." World BankDiscussionPaper 72. World Bank,WashingtonDC. Shleifer, Andrei, andRobertW. Vishny. 1993. "Corruption."Quarterly Journal of Economics 108(3):599-617. Soderbom, Mans, andFrancisTeal. 2003. "AreManufacturingExportsthe Key to Economic Success inAfrica?"Journal of African Economics 12(1):1-29. Stasavage,David. 2002. "PrivateInvestmentandPoliticalUncertainty.''Economics and Politics 14(1):41-63. Streeten,Paul. 1993. "TheSpecialProblemsof SmallCountries." WorldDevelopment 21(2): 197- 202. Tanur,Judith, eds. 1992.QuestionsAbout Questions: Inquiries into the Cognitive Bases of Surveys. New York, NY:RussellSage. WorldBank. 2004a. "InvestmentClimate Assessment: EnterprisePerformanceand GrowthIn Tanzania." World Bank: WashingtonDC. --------,2004b. World Development Report 2005: A Better Investment ClimateFor Everyone. WashingtonDC:World Bank. --------.2005. "Mali: UneEvaluationDuClimatDes Investissements." WorldBank: Washington DC. --------.2007a. "Sub-NationalCostsof DoingBusinessinTanzania: An Assessment of Doing BusinessinArusha, Dodoma,Iringa,Kigoma,Mtwara, Mwanza,Zanzibar." World Bank: WashingtonDC. --------,2007b. WorldDevelopment Indicators. Washington,D.C.: World Bank. ZanzibarInvestmentPromotionAgency. 2004. "ZanzibarInvestmentReport." Zanzibar InvestmentPromotionAgency: Zanzibar. 44 ANNEX 1: SELECTEDEXAMPLESOF SEZDEVELOPMENT Example 1: ZonaAmerica Business and Technology Park, Uruguay ZonaAmerica is one o f the leading-edge free zones oriented to IT, software, regional headquarters, biotechnology and electronics operations. Leading companies include Tata Consulting Services (India) engaged insoftware development for the Spanish-speaking market. Examples o f specialized facilities provided by the zone include: Fiber optic and Wi-Fi network, teleport and microwave links, internet security and on-site help desk, intelligent buildings, wireless perimeter security, research lab facilities, business services center, and medical and daycare facilities. Example 2: Zones Within Zones, the Unique Case o f China special Economic Zones (SEZ's) were established by China to serve as `demonstration areas' for policy reforms and to encourage foreign investment. The economic impact o f these zones has beenfar-reaching, transforming entire regions and economies. The Shenzhen Special Economic Zone (SSEZ) provides a snapshot o f the impact o f the SEZ's on China's economic development. Twenty-three years of growth has transformed Shenzhen from a small sleepy fishing village into a thriving urban metropolis. Today, Shenzhen i s an export-oriented economy that exports about US$48 billion o f goods per year, or 14 percent of the country's exports. The SSEZ has absorbed about $30 billion o f FDIand directly employs about 3 millionworkers. What i s less well known is that the SEZs include hundreds o f other zones. National level zones, all with special and differing incentive regimes include: 14 Open Coastal Cities, 15 Free Trade Zones, 17 EPZs, 54 Economic and Technological Development Zones, 53 High Technology Development Zones, and 15 Border Economic Cooperative Areas. There are many other provincial- andcity-level zones. Example 3: Shannon Free Zone, Ireland The Shannon Free Zone is the world's oldest EPZ, established in 1958. Located at Shannon International Airport, the zone offered investors secure access to European markets, attractive tax benefits and subsidized rent and facilities. Specialized training and manpower development facilities were integrated into zone design from its inception. As a result, export manufacturing activities accelerated. Presently, there are 120 companies employing over 7,500 workers within the zone. As a large share o f the zone's activities are in service sectors, the zone's contribution to overall merchandise exports i s relatively small, accounting for less than 3 percent o f the total. On a yearly basis, zone exports total US$2.5 billion and imports US$1.2 billion. 45 Over time, liberalization o f the Irish economy outside the zone has reduced its relative importance. Nevertheless, the Zone remains an important catalyst for the region, leading the economy's diversification into new, value-added sectors. Example 4: Pomeranian Special Economic Zone, Poland Poland has 14 free zones established throughout the country. Though identifiedas SEZs, the zones generally cover only a limited land area and focus on traditional EPZ and FTZ activities. The program, established in 1995, has been designed as a regional development tool. The experience o f the Pomeranian Special Economic Zone (PSEZ) demonstrates the Polish approach to reusing existing infrastructure for zone development. The PSEZ was established in 2001 as a result o f the merger o f two Special Economic Zones in Tczew and Zarnowiec. The SEZ covers an area o f 348 hectares and is located in the Pomorskie Province, Kwidzyh, Starogard Gdanski, Tczew and Zarnowiec. The Zone will operate untilthe year 2017. One o f the key features o f this zone is its effective use o f existing buildings and infrastructure, and its development o f the grounds o f the former site o f the now defunct nuclear power station project iniarnowiec. Bythe end o f 2000, a total o f 71 permits hadbeen granted to conduct business activities in the Tczew and Zamowiec SEZs. By the end o f 2004 it is anticipated that total investment outlays inthe zone will amount to US$212 million, with at least 4,000 to 6,000 newjobs created. Example 5: The Failed Industrial Linkages Program inthe DominicanRepublic Although many o f the examples discussed above suggest the potential benefits o f SEZs, not all SEZs are as successful. One example o f a less successful zone i s a USAID-sponsored backward linkages program in the Dominican Republic, which illustrates the challenge some countries have experienced indeveloping linkages with EPZs. While feasibility studies revealed abundant EPZ demand for textiles, precision plastic parts, metal stamping, machine shops, and tool, mould anddie making,backward linkages failed to develop. The most important reasons for this include: 0 The relevant sectors frequently did not exist as the Dominican Republic never made significant inroads into the manufacture o f capital andintermediate goods. 0 Local producers generally failed to meet world market standards for price, quality anddelivery terms 0 Local manufacturers often had no interest in supplyingEPZs, being satisfied with current operations andprofitability levels Example 6: Dakar EPZ, a Text-Book Failure Senegal was a pioneer in the creation o f free zones and in establishing its EPZ in 1974. The project generated significant hopes, as Senegal expected to profit from the de-localization o f 46 enterprises from industrialized countries, in the same manner as countries o f the Maghreb, the Caribbean or Southeast Asia had before. The scheme's promoters sought to exploit Senegal's geographical position as well as the port and airport facilities offered byDakar. In1999, almost 20 years after itscreation, Senegal'sauthorities admittedtheproject was a complete failure and ended it.. At the time o f its demise, the Dakar EPZ hosted just 14 active enterprises. The principal reasons for the program's failure were: Excessive bureaucracy involving different institutions inthe country, especially Customs 0 Unnecessarily delays inobtaining the necessary permits, often exceeding one year. Unrealistic goals imposedon potential investors, bothwith regard to jobs creation -each companywasrequiredtoemployatleast150people-- andtothesizeof the initial investment 0 Poor reputationo f the local workforce, judged to be unproductive and expensive 0 Elevatedcosts for other ingredients o fproduction such as energy, water, and communications. Rigidlabor regulations-employment contracts were permanent and employers didnot have complete freedom to recruit the people they wantedneeded. Example 7: Private Free Zone Development inthe Dominican Republic The country's 22 public zones were established primarily as a means to encourage regional development outside the capital area. Instead, the private sector zones, which today number 31 (including joint public-private ownership), are heavily concentrated around Santo Domingo, The Dominican Republic's largest population center and close to critical port and airport infrastructure. There are currently 194 companies operating in the public zones and 326 inprivate orjoint ownership zones. Surveys o f zone enterprises highlight the role o f the private sector in upgrading the facilities and services required o f export enterprises, particularly those in manufacturing (Rhee andBelot, 1990). The private zones, driven bymarket forces, are locatedprimarilyinthe vicinity o f Santo Domingo, providing access to the country's highly qualified andproductive labor force, as well as access to highquality transportation infrastructure, Most importantly, zone enterprises here demonstrate a willingness to pay higher prices for their space (in some cases, up to three times higher) inreturn for high quality infrastructure facilities, and services. The private zones boast state-of-the-art telecommunications services, well-developed business support services, andquality manufacturing andoffice space. Source: Gauthier (2004) 47 zx W Bm 4- E ?- v: Ee! 0I: vl L 0 v, c I 8Y 0 .ev) tz 8x b 0 0"c I tz R 3 W m C II c. b Y v, M 3 ri ri; KMa m 0 81 -28a > 3 1 @ rr 0 2b s* N- 0 0 28 N e, c m 1 I I I 5E 'I 1 Bb 5 0 v) a * I- .e E 6 0 k -0 8 e, 'C B Y m *%3 n $ * 0 E r 0 8x 3 5 a, N W 5 1 d 2 t! s1 & w D 4m 8 e, > G - N 3 P, i E.- E.- m m 5W q M k .C 0 s x & 0 Y Y 0 c .-88 I .I -5 B c! 5e, W W e. t: 8 .-84 Y -0 --"! Y 80 o! 1 8E - c " AY 5 Q 9 3 0 n + z s sI s -0- w s t i I I I n - U 3 0 6 !? Y .-8mc rd Y N .-E 5 Q Q- 0 2 i2 1 kx E 0 3 c 8x W c 0 * n 0 a, E E 3E 0 5 8 8 I T 10; 7- I B L; 0 3 e R a k -E$ 0 3 -3 R ._ P2s W * ae! e 0 W a 0 R C h E -0 0 e, axx *0 C -0 0 c 8 a 3 h8 !i 0 4- R ii e! -4sa '3 E C E 3 '3 * E 50 cd e. 5 0 x 0 z 5 .e 2 VY $Z 3b '"v ) W * 22 aaR D 3 d sB 3 I; B 3 rA B Y E BQ 1 1 /II I O1 -0 w * .I Q z .I 8 P f I e, 5 2 -s.- * I 8 e, E vid se, rr 0 i2 24E vi+ 8Eax +? s5 E E ox c I e'pj nrl v) 0 c) H ."*m zme,>0B G & R li 1 A b za k 0 23V 8 e. 'EEE3 x Y 8 E G cb 5 5 '5 c 3 E .- c ii m [I 3 R h .-m 0 I P 00 3 00 0 00 3 11, Il o o I I I l l .i- W d I I I 'B -2 l o o o o E .C 00 N TI x - rc e, * 0 * 3 c g t 1 c 110010 d d d d m m m m N N N N 3 e e 3 * 2III e 8 II I l l W 00 d - n d 9E 00 00 d ed d x w 3 QI , rn . . At4 x CODE FORQ. 28 1...Healthcare facilities operatedbythe fm 2.. .Private healthproviders (including traditional healers) 3.. .Public facilities 4.. .Facilities o fnon-profit or charitable organizations 5.. .Other (specify inchart) CODEFOR Q.29 1...No significant out-of-pocket expenses necessarysince treatment is free or low cost 2.. .High out-of-pocket expenses but reimbursed by employer fully or partially 3.. .High out-of-pocket expenses but reimbursed by insurance company fully or partially 4.. .High out-of-pocket expensesbut financial support from frienddfamily not belonging to myhousehold 5.. .High out-of-pocket expensesborne by my household 6...Other (specify inchart) 94 1 Data on total firms from 2001 Central Register o f Enterprises as described in Office o f Chief Government Statistician(2005). 2 Results inthe previous report were presented in 2002 US dollars. For consistency with results for more recent surveys, these have been converted into 2005 dollars. Results are qualitatively similar in2002 US dollars. 3 4 Although for a variety o f reasons, critics often claim that perception-based data provide little information on the investment climate, recent studies show that perception-based measuresperform well both incross-country analysis and inwithin country, fmlevel comparisons (Gelb and others, 2006) 5 Inaddition to size and remoteness, Briguglio (1995) notes that vulnerability to natural disasters and environmental factors might also be important. These issues are not discussed here because they are not easily addressed using investmentclimate data. 6 Easterly and Kraay (2000) conclude that this is because these economies are more open to trade 7 A related problem i s that small countries have a smaller labor pool from which they can draw public administrators (Briguglio, 1995). Since it is costly to train administrators who need specialized skills in small countries, it might be necessary to train people abroad for these tasks. However, given that small economies appear to be especially prone to outward migration (Dommen, 1980), this can be costly if the trained individuals do not return (Briguglio, 1995). 8 Data for Zanzibar are from Office o f Chief Government Statistician (2007) and data from Dar es Salaam are from National Bureau o f Statistics (2006). 9 Several empirical papers looking at Tanzania and other countries in Sub-Saharan Africa inthe last decade found that small firms are less likely to export than are larger firms. Most notably, Grenier and others (1999) found that large Tanzanian enterprises export more than smaller enterprises. Using data from several countries insub-Saharan Africa from the mid-l990s, Bigsten and others (2004), Soderbom and Teal (2003) and Clarke (2005) noted similar results. 10 See Eifert, Gelb and Ramachandran (2005), World Bank (2004a) and World Bank (2005) for comparisons o f productivity inTanzania with other countries inAfrica and high-growth developing economies. 11 See World Bank (2004b), Dollar, Hallward-Driemeier and Mengistae (2003) and Escribano and Guasch (2005). 12 See Eifert, Gelb and Ramachandran (2005). 13 See, for example, Bertrand and Mullainathan (2001); Recanatini, Wallsten andXu (2000); and Tanur (1992). 14 The difference is also close to significant after controlling more fully for fm size. See regression in the Appendix. 15 World Bank (2004a) discusses some recent changes intax administration inTanzania. 16 Inregressions similar to those shown for macroeconomic stability inthe Appendix, ina regression with a dummy variable indicating that the fm saw tax administration as a serious obstacle, the coefficient on the year dummy i s statistically insignificant inthe regressionsboth with and without the other controls. 17 This difference i s statistically significant 18 This difference i s statistically significant 19 See, for example, Djankov and others (2002) and Shleifer and Vishny (1993) 20 The difference i s statistically significant at a 10 percent level before controlling for size and is just statistically insignificant at a 10percent level after (see Appendix). 21 This is noted as one o f the reasons why private borrowers do not borrow from banks in Zanzibar's Growth Strategy (2006-20 15) (Ministry o f Finance and Economic Affairs, 2006). 22 To be classified as not wanting or needing a loan, the manager had to reply `yes' to either that they didnot want a loan or that they didnot want to incur debt. 95