Location , Ownership , Origin and the Spillover-Productivity Nexus . Evidence from Uganda Manufacturing Firms

The main purpose of this article is to investigate the drivers of labor productivity in the firms at the intra -industry level with focus on the spillover effects of FDI. Using a fixed effects approach, we estimate an expanded Cobb-Douglas production function in its intensive form to isolate the effects of increased capital intensity on labor productivity as well as the spillovers, using annual Private Sector Investment Survey data collected on the Ugandan manufacturing firms over the period 20072010. Over all, there are significant negative horizontal spillovers for the domestic firms in Uganda, with OECD-originating FDI appearing to be the main source of such effects. By location, these are most adverse in the western and eastern regions and better spillo vers can be traced in the central region. Additional findings point to firm size, labor quality and profit as positive contributors to labor p roductivity, whereas technology gap exhib its a detrimental impact just as we document no significant effect of cap ital intensity. Larger domestic firms appear to benefit significantly from spillovers in industries where foreign firms have a larger p resence. The aforementioned findings reflect the need for well-designed policies to improve the competitiveness of local firms particu larly v ia an incentive-equal opportunity-policy that captures both domestic and foreign investors and to improve in frastructure and other investor-friendly environment in the East and Western parts of Uganda. Similarly, our results suggest tha t the promotion of joint ventures (foreign) is likely to generate unequivocal benefits to the manufacturing sector in Uganda not only in terms of less negative horizontal spillovers but also from the labor quality, firm size and profit spillovers perspective. Finally, the finding of learning difficu lties of domestic firms from foreign firms calls fo r programs in line with skill acquisition through job train ing and the review of the curricu lum to focus on labor quality.


Introduction
Foreign d irect investment (FDI) is beco ming a very important source of foreign capital fo r many economies.On this basis, many African countries have embraced polic ies to provide a friendly environ ment for foreign investment, sometimes at the expense of domestic investment.For examp le, the removal or relaxation of FDI restrictions and the provision of foreigners with incentive packages in the form of subsidies on in frastructure, lower taxes, tax holidays, free land, and import duty exemptions inter alia.The justification often advanced for such policies hinges on the argument that FDI assists human capital format ion, imp roves management skills, contributes to international trade integration, helps create a more competit ive business environment, enhances enterprise development and employ ment creation.While an opposite phenomenon is equally likely, particularly in terms of the crowd ing-out effect, perusal in the available records seems to suggest that these policies might be instrumental not only in catalyzing economic growth, which is the most potent tool for poverty alleviation (Blo mström et al., 2000) but more so in accelerating FDI inflows to developing countries i f any development so far is to be sustainable.According to the UNCTAD (2015), inward FDI flows to developing economies in 2014 reached their highest level at $681 billion with a 2% rise, reflect ing an extended lead in global inflo ws.For Sub-Saharan Africa in particular, an increase of 5% to $42 billion is recorded while the entire Africa is said to have contributed a meager 3% of the world FDI inflo ws in 2007, but which shot to 5.3% in 2010.Elsewhere, The State of East Africa 2013 report identifies Uganda and Tanzania to have attracted the most FDI in 2012 in the East African region.Net FDI inflo ws to Uganda in part icular increased on average fro m $90.64M to $242.71M in the periods 1991-2000 and 2001-2005, respectively, reflecting a 168% increase.For the year 2006 and 2009, an increase of 24% fro m $644.26M to $798.77M is recorded (UNCTA D, 2010).Interestingly, a previous study by the World Bank (2004) found Uganda to be 60% and 40% less than Kenya and Tanzania respectively in labor productivity.Mindful of such findings, in the current study we investigate whether or not domestic firms in Uganda benefit fro m the observed inflows of foreign investment particularly in the manufacturing sector and the relevant channels via which these spillovers could be re aped.
According to Blo mström et al. (2000), FDI to developing countries might trigger technology spillovers to low technology domestic firms.A related argu ment had earlier on been advanced in a seminal work by Ait ken and Harrison (1999) who points out that domestic firms could gain fro m foreign firms via accelerated diffusion of new technology once foreign firms introduce new products or processes to the domestic market.On the other hand, by simply observing nearby foreign firms or when domestic emp loyees move from foreign to domestic firms, do mestic firms could increase their p roductivity through technology diffusion.These arguments notwithstanding, the presence of foreign firms may also lead to negative spillover effects on the host -country firms.For e xample, the latter may not be able to compete favorably with foreign firms in the labor markets and as a result, there is likely to be increased mobility of skilled workers fro m do mestic to foreign firms under the assumption that foreign firms pay more than domestic firms.In addit ion, as indicated by Aitken and Harrison (1999), entry of foreign firms is likely to disturb the existing market equilibriu m in the host country thereby forcing do mestic firms to produce less output and increase their average costs.Failure to meet these costs may force some do mestic firms to exit production and in the long run p roductivity will be reduced.It is therefore not illogical to hypothesize that competition may result into reduced market shares, increased poaching of bes t workers fro m do mestic firms and making access to credit mo re difficult for do mestic firms because foreign investors may be lower-risk borro wers compared to their local co mpetitors.The latter scenario may force domestic firms to produce less output.
Given the two opposing theoretical justifications above, economists and other scholars still grapple with the puzzle regarding the ro le of Multi-national corporations (MNCs) in local firms.In the current paper we are interested in the contribution of FDI to the p roductivity of do mestic manufacturing firms in Uganda.The choice for the manufacturing sector is precipitated by a reasonable argument that the industry has the potential to be a driving force and cornerstone for Uganda's modernization and job creat io n.Yet its share in aggregate output has consistently remained low despite the macroeconomic and industrial policies to improve its competitiveness.At the same time, both foreign and domestic firms have registered tremendous increase at least in number (U BOS, 2014).Perhaps one would expect to observe a sizeable contribution of the manufacturing sector as domestic firms accelerate production, imp rove productivity and efficiency by probably learning fro m foreign firms and gaining fro m technological diffusion.One vital fact however is that the manufacturing sector in Uganda, well-known to be do minated by s mall and mediu m enterprises (SM Es), is still largely engaged in the production of low value added goods, comprising basic consumer goods, processed foods, tobacco and beverages, non-metallic minerals and metallic fabricat ion, wood and wood products, chemicals and chemical products, leather and footwear, textile and wearing apparels, and sawmilling, printing, and publishing.Heavy investments by foreign co mpanies are more p ronounced in textiles, steel mills, tanneries, bottling and brewing, and cement production, (African Development Bank (AfDB), 2014).Production has been growing in varying degrees fro m year to year, as more and more do mestic and MNCs continu e to join the sector.For example, evidence fro m UBOS (2014) shows that overall, the index o f production for manufacturing for the year 2013 was 199.5 demonstrating a 3.2%increase co mpared to the year ending 2012.Growth was highest in 'Food Processing' with 10.5% growth, followed by 'Saw Milling' (8.4%), 'Metal Products' (6.2%) and 'Bricks and Cement' (5.0%) while other industry groups combined recorded a positive rise of 5.7%.Subsectors whose output reduced less than zero since the year, 2005 were 'Drinks and Tobacco' (-2.0%), 'Chemicals; and foam products' (-0.2) and Textiles (27.7).Specifically, Text iles, Clothing and Footwear dropped by 27.7%as shown in Table 1.The yet unanswered question however is whether the presence of foreign firms has had any significant impact on the production capacity of domestic firms as the manufacturing sector in Uganda continues to exhibit gro wth in average production.The current paper bridges the gap by examining the impact of FDI on labor productivity over the period 2007-2010, a period that coincides with significant FDI inflo ws particularly in the manufacturing sector.The relevance of our investigation is not unambiguous.Scanty literature exists in this area with a few focusing mainly on investigating the determinants of FDI and impact on econo mic growth (e.g.Obwona, 2001) while others that link FDI spillovers to domestic firms in the LDCs have used macro data (e.g.Karbasi et al., 2000;and, Mutenyo et al., 2008).The results fro m the latter strand of literature might obscure individual firm characteristics and lead to irrelevant policies.To overco me this problem, other studies use mic ro level data to analyze individual countries (e.g.Gorg and Strobl, 2005;Annika, 2006;Galeotti, 2009;and, Waldkirch and Ofosu, 2010).The underlying argument for a micro approach is that the conclusions are dependent on the particular country under analysis.To our knowledge, no similar analysis using firm-level data has been done for the case of Uganda.In addition to the scarcity of literature, the existing empirical evidence on such spillover effects fro m FDI is to-date mixed and ambiguous.While some studies document significant and positive spillovers fro m FDI (e.g.Javorcik, 2004;and, Nguyen 2008), others register a negative effect (e.g.Konings, 2001;Blo mström et al. 2000;De Backer and Sleuwaegen, 2003;and, Bwalya, 2006).Moreover it is also equally not uncommon to find records of no significant spillover effects fro m fo reign firms on domestic firms (e.g.Kathuria, 2001;and, Mebratie and Bed i, 2013).
In light of these considerations, our contribution to the existing literature h inges on the an alysis of FDI spillovers on Ugandan manufacturing firms.For, the co-existence of domestic and foreign firms is an unquestionable phenomenon in this region and Uganda in particular.Specifically, we focus on three main questions: First, what drives labor productivity in the manufacturing firms in Uganda?Does foreign presence matter?Second, which subsectors benefit more than others in the presence of foreign firms?Finally, does the origin of foreign investment, the geographical location in the recipient country and ownership influence the nature of spillovers to domestic firms?The essence of the last multifaceted objective is to trace the impact of heterogeneity of FDI inflows on domestic firms in Uganda via spillovers.It needs not to be overemphasized t hat the origin of the foreign firms in Uganda is mult ifo ld; while some co me fro m the Organizat ion for Econo mic Co -operation and Develop ment (OECD) countries, the others are from Asia and other African countries (see Table 2 below).Given the increasing presence of Asian countries, and China in particular, in Africa, we wish to establish the nature of spillovers drawn fro m each source.The focus is on the role of OECD countries vis -à -vis other p ro minent but growing sources of inward FDI to Uganda, viz.Asia and Africa.Similarly, we uncover the impact of the 'geography of spillovers' within the host country.The results from our study suggest an existence of negative horizontal spillovers.While firms located in the central reg ion experience greater spillovers in relat ion to others in the east or western parts of Uganda, the foreign investors originating fro m OECD countries have a remarkably better contribution in terms of spillovers to the Ugandan domestic manufacturing firms in co mparison to those sourced from Asia or other parts of Africa.Additional evidence exhib its larger technology gaps as detrimental to labor produ ctivity wh ile the opposite holds for pro fit and labor quality.Incidentally, we fail to find significant ev idence of the role of capital intensity in the facilitation of the emp loyee productivity.
The rest of the paper is organized as follows: While in Sect ions 1 and 2 we present the introduction and a review of the relevant literature respectively, Sect ion 3 bears an explanation of the methodology and data.The results and discussion are then covered in Section 4, while we provide the summary and concludin g remarks in Sect ion 5.

Literature Review
According to theory, there are t wo effects of FDI, namely : the direct and indirect effects.Directly, FDI could increase emp loy ment opportunities, capital accumu lation, international markets through exports, usage of advanced equipment and technology (Blo mström et al. 2000).The indirect effects of FDI are spread through specific contacts between foreign firms and domestic firms.When MNCs penetrate new markets, they always protect themselves against their domestic competitors by preventing technology leaka ges and other spillovers fro m taking place.The protection channels include inter alia intellectual property rights, paying higher wages to prevent the movement of labor fro m one firm to another and moving to countries that cannot imitate their proprietary assets.However despite these measures, some indirect effects are diffused to their do mestic competitors.In the current study, we focus on the indirect effects of FDI associated with technology spillovers fro m M NCs to manufacturing firms.Do mestic firms may indirectly benefit fro m the presence of foreign firms in a country through both vertical (inter-industry) and horizontal (intra-industry) spillovers.The vertical spillovers are known to occur as a result of the interaction between foreign and do mestic firms not in the same industry.On the other hand, domestic firms are likely to benefit fro m the horizontal spillovers that work mainly among firms within the same industries through three distinct theoretical channels.The first channel is where technolo gy diffuses to domestic firms of the hosting countries through demonstration and imitation effects.Th is takes place when domestic firms observe the actions of foreign firms, learn about new technologies which they can apply locally and imp lement the techniques to increase domestic productivity (Ait ken et al. 1997).In fact after observing the product innovation fro m M NCs, do mestic firms may learn the successes and failures of foreign firms which may help them to increase their productivity.The second chan nel is the competition effect that is also related to the demonstration effect but helps domestic firms to maintain their market shares by improving the product quality and may encourage them to operate more efficiently and adopt new technologies than it would have been the case without MNCs (Bit zer and Gorg, 2008).However, if the co mpetition effects dominate, the effects of FDI on domestic firms could be negative.Aitken and Harrison (1999) argues that, the presence of FDI in a host country destabilizes the market equilibriu m forcing domestic firms to produce less output and reduce their market shares.In addition, competition may also force domestic firms to reduce their production below optimal levels and this increases the prices of their products.The third channel is that of the labor turnover or mobility effects where by workers and managers who received training and were originally emp loyed by the foreign firms may either move to do mestic firms with their acquired skills fro m foreign firms or establish their own businesses in similar fields taking with them their upgraded human capital which thus becomes available to do mestic firms, raising their measured productivity.Moreover, as argued in Haddad and Harrison (1993), the scope of technology diffusion and transfer depends on the absorptive capability of domestic firms.
Intuitively, an increase in the difference in technological complexity between foreign and domestic firms theoretically discourages any gain fro m spillovers by domestic firms.
In light of the above theoretical underpinnings, there is a continued debate concerning the direct and indirect effects of FDI on domestic firms.Several emp irical studies have addressed the issue but the findings are overly inconclusive and characterized by a mixture of evidence.A study by Javorcik (2004) for examp le looks fo r both horizontal and vertical spillovers using firm-level panel data and finds that total factor productivity (TFP) of Lithuanian firms is positively correlated with the extent of potential c ontacts with MNC customers in downstream sectors.The author indicates that a one standard deviation increase in foreign presence in the buying sectors is associated with a 15% increase in the productivity of firms in the supplying industry.The observatio n of the positive productivity effects however is only valid for joint ventures and not the fully o wned foreign affiliates.Relatedly, Marcella and Resmin i (2010) investigate both the horizontal and vertical spillovers in Bulgaria, Poland and Roman ia and document a positive relationship between FDI and the domestic firms but conditional on the absorptive capacity and technological levels of the host country.Similarly, Nguyen (2008) examines the effects of FDI on the domestic productivity in Vietnamese manu facturing industries through horizontal and vertical linkages and records positive spillover effects fro m FDI through both horizontal and backward linkages.In addit ion, the study notes that the Vietnamese regions benefit fro m FDI spillovers but these spillovers differ fro m region to region; and that the private firms have strong linkages through technical assistance and technology transfer with foreign invested firms wh ile the lin kages of state owned enterprises with foreign invested firms are very weak.Do mestic firms with higher human cap ital stock, better financial development and lower technology gap are found to experience positive effects fro m FDI resulting into increased productivity.On his part, Bwalya (2006) examines the productivity spillovers from foreign firms to domestic firms using firm level data for manufacturing firms in Zamb ia emp loying a Cobb -Douglas production function.The GMM estimat ion results therefrom indicate that the productivity of domestic firms reduces as foreign presence in the sector increases which the author attributes to adverse competition effects of inward FDI.
On the other hand, an adverse effect of foreign presence on domestic firm productivity is not uncommon in emp irical literature.Fo r example, Wald kirch and Ofosu (2010) examine the foreign presence, spillovers and productivity in Ghanaian manufacturing firms and find that foreign presence in the manufacturing sector has a negative effect on the labor productivity of domestically owned firms.The authors argue that d omestic firms experience negative effects (intra-industry effect) because foreign firms in Ghana not only get incentives but also face a lo wer marginal cost which allows them to co mpete more successfully in product and factor markets and attract more skilled workers fro m the domestic firms.They argue that while co mpetitive pressures in the long run may induce efficiency through knowledge spillovers, in the short run FDI inflows may be associated with negative consequences for do mestic firms.Likewise, Ping et al. (2009) use a large panel data set covering all manufacturing firms in China over the period 1998-2005 to examine whether there are some productivity spillovers fro m FDI to domestic firms.In estimat ing the productivity, they control for a possible simu ltaneity bias using semi-parametric estimation techniques and find that Hong Kong, Macao and Taiwan (HMT) invested firms generate negative horizontal spillovers wh ile Non -HMT foreign invested firms (mostly fro m OECD countries) are associated with positive horizontal spillovers in China.However these two opposing horizontal effects cancel out at the aggregate level.Similarly, a study by Galeotti (2009) on the Czech Republic, using panel data over the period 1990-2006, confirms the evidence that foreign direct investors produce negative spillovers on domestic firms.The findings reveal that these spillovers do not play a dominant role for the performance of privatized do mestic firms in the glass sector and the importance of taking into account the indust rial sector in the study of spillovers.However, addit ional findings support the importance of geographic pro ximity and the agglomerat ion of foreign direct investors as a channel of spillovers.
A recent study by Mebratie and Bedi (2013) however investigates the relationship between FDI and labor productivity of domestic firms in South Africa using two -period (2003 and 2007) firm level panel data and find no spillover of foreign firms on labor productivity.Similar findings are recorded in Chang -Tai (2006) who, using a unique plant-level panel dataset fro m the Chinese manufacturing sector to measure the effects of foreign firms on the productivity of domestic firms, reveals that foreign firms are more productive than domestic firms in the same industry but that there are no spill-over effects fro m foreign to domestic firms.Support for such an outcome can also be traced in Ruane and Ugur (2004) who investigate the spillover effects of FDI on the labor productivity levels of domestic firms in the Irish manufacturing sector for the period 1991-1997.Controlling for capital intensity and labor quality of these plants, they find no evidence of significant productivity spillovers fro m FDI.Gorodnichenko et al. (2007) reiterates similar findings in their analysis of the spillovers in 17 emerg ing countries and find no horizontal spillovers, except for older firms and firms in the service sector.Indeed Annika (2006) evaluates the relationship between FDI and Indian pharmaceutical industry and reveals that there is no correlation between FDI and productivity in do mestic firms because of the "absorptive capability".
In some other literature however, the results are dependent on the country under analysis.For example in Vahter (2004) while FDI is negatively related to domestic productivity in Estonia, it is positively related in Slovenia.
Moreover, the effects appear to be dependent on the conditions in the recipient countries such as types of ownership (jo int venture or fully foreign owned) and the level of skilled labor.Gorg and Strobl (2005) look at the issue of ownership but from a different perspective.The authors employ Ghanaian data to investigate whether or not the owner of a domestic firm had previous experience with a foreign firm wh ich they relate to firm-level productivity.Their study findings suggest that firms run by the owners who worked for foreign firms in the same industry before opening up their own firms, are more productive compared to other domestic firms that did not have contacts with foreign firms.Viewed fro m technical efficiency, however, Faruq and Yi (2010), using the Data Envelop ment Analysis (DEA) technique to estimate the technical efficiency of firms in Ghana across six manufacturing industries during 1991-2002, observe that the characteris tics of manufacturing firms in Ghana, viz., firm size, age, foreign ownership, and the mix of labor and capital used during the production process, are positively associated with firm efficiency.On his part however, Kolasa (2007) examines the existence of externalities associated with FDI in a host country using firm level panel data for the Polish corporate sector and finds that while do mestic firms benefit fro m foreign presence in the same industry and in downstream industries, the absorptive capacity of domestic firms is highly relevant to the size of spillovers.Aysa (2011) evaluates the latter argument in the analysis of the horizontal productivity spillover effects of foreign ownership on Turkish firms by emp loying a panel of 215 firms over the period 2004-2008.The study findings however reveal that although domestic firms benefit fro m the productivity spillovers fro m foreign -owned firms, absorptive capacity does not matter for productivity spillover benefits.Instead, Sinani and Meyer (2004), who emp loy a panel data technique in Estonia for the period 1994-1999 to study technology transfers from FDI, argue that the magnitude of the spillover effect depend on the domestic firm's size, its ownership structure and trade orientation.
With regard to origin, Monastiriotis & Bo rrell (2013), using firm-level data fro m the Business Environ ment and Enterprise Performance Survey (BEEPS) covering 28 transition countries over the period 2002-2009, estimate the direct and intra-industry productivity effects of foreign ownership and examine how these differ across regional b locks, viz, Central and Eastern Europe (CEE), South and Eastern Europe (SEE), and, European Neighborhood Policy (ENP), according to the origin of the foreign investor (European Union -EU versus non-EU), across geographical scales (pure industry versus regional spillovers) and for different types of locations (capital-city reg ions versus the rest).Their results suggest that FDI of EU origin plays a distinctive role in the countries concerned by raising domestic productivity significantly more than investments from outside the EU.
Overall, the empirical ev idence on the productivity spill-overs is mixed, with some studies suggesting positive spill-overs effects while others either find negative effects or no spill-overs at all.The characteristics of firms and Heterogeneity appear to play a key role in explaining these differences.A country analysis is therefore a more prudent approach than a cross -country assessment and is therefore preferred in the current study to avoid irrelevant policies.The quantificat ion of the spillover effects is still a missing link for the case of Uganda.We focus on the horizontal spillovers in intra-industry to shed light into the association between labor productivity and firm characteristics.Moreover issues regarding the location of the firm, profits, and an analysis to capture the ownership and sector-wide effects of spillovers, in addition to the interactive effect of the trad itional variables in the labor productivity model such as capital intensity, technology gap, firm size and profit, appear to have escaped the attention of many prev ious studies.We contribute to literature by taking all these missing lin ks into consideration.An additional novelty in the curren t analysis is the evaluation of the role of firms orig inating fro m OECD countries vis -a-v is other regions, particularly Asia.Specifically, fro m which foreign investors (by origin) do domestic manufacturing firms in Uganda gain most?We equally endeavor to provide quantitative evidence to this informat ive question.

Model Specification for Spillover Effects o f FDI
Emp irical studies investigating the technological spillovers fro m FDI to domestic firms have either used total factor productivity (TFP) or labor productivity as dependent variables.Since our interest is to isolate the effects of increased capital intensity on labor productivity and to compare our results with prev ious studies that examine the impact of FDI on the domestic firms using the same measure (e.g.Jordaan, 2008, and, Mebratie andBed i, 2013), labor productivity is preferred to total factor productivity or any other alternative measures.
The model starts by assuming a Cobb-Douglas production function: where Y stands for output; A stands for the efficiency parameter; K and L represents the factor inputs used in the production process (capital and labor, respectively) while α and β denote the elasticity of capital and labor inputs, respectively.Assuming there are constant returns to scale, such that α + β = 1 (2a) Following Jordaan (2008), we assume the possibility of not only constant returns but also increasing returns or decreasing returns, imp lying that equation (2a) can now be exp ressed as α + β = δ , (2b) where δ can be smaller, equal or greater than 1.Th is is the same as β =δ −α .
Equation 2(b) can likewise be expressed as: (2c) Substituting exp ression 2(c) in (1) we get: We then divide both sides by L to obtain a production function in its intensive form as: Suppose we include the firms (i) and time (t ) element in equation ( 4) and then take the natural logarith ms on both sides, we obtain a linear function given by: Equation ( 5) contains a parameter (δ −1) that is unknown; we use this parameter to construct a proxy to control for scale economies that is used in our specified model; the variable is included to control for increasing returns to scale.Hence we can say that labor productivity depends on the productivity efficiency (A), which is a constant, capital intensity (KI), labor quality (LQ), firm size (SIZE) and fo reign presence (HORI).Therefore our preferred model is specified as: Here LP is domestic labor productivity, i  stands for the error term for the ith firm, u is the unknown individual firm effects, v error term with zero mean and constant variance. stands for the coefficients to be estimated, i is the firm while t stands for time.All variab les are measured in nominal terms and are estimated in log forms.We include additional variab les, X, which may affect labor productivity.These include exports, imports, profits and technology gap.However, we leave out the first two due to data constraints.
In order to find out whether the spillover effects depend on the size of foreign firms, capital intensity or technological gap, we introduce interaction terms as follows: In the presence of firm size (SIZE): Then the marginal effect is equivalent to: In the presence of capital intensity (KI): The marginal impact in this case will be: In case of technology gap (TG): With regard to profit: The corresponding marg inal impact is then: The study emp loys fixed effects approach.However we also run regressions using the alternative random effects and pooled OLS, but due to space the results therefrom are not presented since they appear not to be significantly different fro m those of the FE estimator and are available on request.

Definition and Measurement of the Variables
The dependent variable is labor productivity (LP) measured as the ratio of the firm level turnover (Y) to the total number of employees (L) in the Ugandan manufacturing sector; L is divided into skilled and unskilled.The independent variables of interest include capital intensity (KI -measured as the ratio of fixed assets (K) to the number of emp loyees in each firm), labor quality (LQ -measured as ratio of skilled to unskilled labor), firm size (SIZE -the ratio of the firm output to output for the largest firm in each sector), technology gap (TG -the ratio of average productivity of the foreign firm to the average productivity of the domestic firm in the same industry), profits and spillovers (HORI).A positive sign of the coefficient of LQ is expected and contributes to an increase in labor p roductivity.The size of the TG might be positively (learning potential) or negatively (learning difficulty) related to spillovers.We also expect positive signs on the KI and SIZE coefficients.We capture the impac t of foreign presence on domestic firms in the same industry (spillover effects) using HORI measured by the share of foreign emp loy ment to total industry emp loyment (as in Blo mström et al. 2000).It is however difficu lt to state a priori the d irect ion of the impact of foreign presence as imp lied in theory.Tab le 3 displays the descriptive statistics and the pairwise correlation mat rix of the variables.

Data Source and Estimation Methods
The firm level data used in this study was obtained fro m the annual Private Sector Investment Survey (PSIS) jointly conducted by the Bank of Uganda (BOU) and the Uganda Investment Auth ority (UIA).It is the best detailed dataset available covering the period of four years (2007)(2008)(2009)(2010) for Ugandan manufacturing firms.The objectives of the surveys are: (i) to enrich and update the national database with vital info rmation on private entities to guide planning and policy ((ii) to identify gaps to facilitate provision of better services that may attract more foreign investors and (iii) to assess the investment promotion strategy effectiveness and the investment climate in Uganda (see, private sector investment survey 2011).The data set contains information on location by region, ownership either foreign or do mestic, emp loyment levels and type or characteristics of the emp loyees (i.e. both skilled and unskilled labor), value of fixed assets, total output (proxied by turnover) and industries.The key piece of informat ion provided in the data is the ownership of firms.Specifically, ownership is divided into two: domestic and foreign ownership and we use this information to construct the horizont al spillovers.A summary of the data by subsector, orig in, location and ownership is presented in Table 4 to augment Table 2.In Table 6 we prov ide additional direct effects regarding the spillovers on labor productivity by subsector of manufacturing (Co lu mns 1-7) and by ownership (Colu mns 8-12).One outstanding result is that the pharmaceuticals subsector exh ibits positive spillovers.While we take this outcome with a grain of salt g iven the few observations in the relevant subsample, there is a possibility of the learn ing potential with do mestic firms positioned to gain fro m foreign firms.However, in the same subsector the larger the size of the firm the smaller the advantages accrued to it in terms of productivity.While it is recorded that capital intensity is inversely associated with productivity in the Pharmacy subsector, the Metal subsector appears to gain out of an increase in the same.Similarly Profit matters for the latter just as the opposite is true for the fo rmer.And while labor quality is significantly helpful in the Paper and Pharmacy subsectors, it is deleterious in th e Rubber and Plastics subsectors.On the other hand, the same variable matters significantly for joint ventures that are predominantly foreign as opposed to any other type of ownership.A similar observation applies to profit.Specifically, a 1% increase in profit levels appears to increase labor productivity by about 0.018 percentage points.However, considering the same type of ownership, capital intensity appears to impact negatively on productivity, though the result is weakly significant and the coefficient magnitude small (-0.008).There is no evidence for the other types of ownership regarding the effect of the same on labor productivity, even though the possibility of the positive influence in Joint Venture Local and in firms that are Wholly Foreign if we go by the economic sign is not dismissible.Overall, Joint Ventures have the potential to gain in labor productivity fro m an imp rovement in capital intensity, labor quality, profits and firm size.Besides the direct effects of fo reign firms on the Ugandan manufacturing sector, the study also examines indirect channels through which FDI may impact labor productivity of domestic firms.In Specification (2), Table 5, an interaction of foreign presence (HORI) with the technological gap (TG) provides evidence that as the technology gap increases by 1%, the effect of horizontal spillovers on labor productivity increases by about 0.047 percentage points.Note that the total marg inal impact is -0.93 (i.e.-0.738+0.047*-4.07),which could suggest that in the presence of large technology gaps, horizontal spillovers would negatively be associated with labor productivity.For, domestic firms in such an environment appear to have learning d ifficult ies fro m foreign firms.
The results obtained are similar to those of Savuth (2012).
In addition the study also examines the indirect impact of foreign presence (HORI) with cap ital intensity (KI) and the firm size (SIZE) as shown in Table 5, Specifications 3 and 4, respect ively.The interaction between HORI and capital intensity (HORI*KI) reveals that the differences in capital between foreign firms and do mestic firms did not have any influence on the productivity of domestic firms.On the other hand, the interaction of HORI with SIZE (HORI*SIZE), intended to examine whether domestic firms experience the scale economies, reveals, as shown in Co lu mn (4), that firms o f a relatively larger size benefit better than smaller size firms in an industry with a strong presence of fore ign firms.With a 1% increase in the foreign-firm presence, domestic firms' productivity, given a med ian level of firm size, reduces by 0.88.Th is reduction is lower than when the median level of technology gap is under consideration but higher in cases wh ere either capital intensity or profits are interacted with spillovers.
As previously indicated, the current study extends the analysis by investigating the hypothesis that the origin of the investor may play a particu larly relevant role in the realizat ion of horizontal spillovers in Uganda.Specifically we address ourselves to the follo wing question: Do OECD -originating FDI have a productivity advantage over other regions in terms of spillovers to domestic firms in Uganda?In line with our prior expectation, the answer is in the affirmat ive: Foreign investors originating fro m OECD countries have a remarkably better contribution in terms of spillovers to the Ugandan domestic manufacturing firms, in comparison to those originating fro m Asia or other parts of Africa.In fact, going by the results in Table 7, Colu mn (4), it is visibly evident that when foreign investors are from OECD reg ion, the unique effect of HORI on labor productivity is -0.561 (i.e.-0.898+0.337(1))co mpared to a worse scenario when FDI is fro m Asia, and in the latter case the unique effect is -1.01.
In a Table 7, Colu mns (1) to (3), we also present the locational advantage and disadvantage of firms in the spillover-productivity nexus.Though weakly significant at 10% statistical level, firms located in the Central region enjoy higher productivity resulting fro m horizontal spillovers.The relevant coefficient suggests that domestic firms' productivity, given the central region to be the location of firms, decreases by 0.76 if foreignfirm presence increases by 1%.On the other hand, the point estimates in Colu mn (3) suggest that domestic firms' labor productivity, given that the manufacturing firms are located in the western reg ion, reduces by 1.73 if foreign-firm presence increases by 1%.The reduction is however lower if the firms are located in the Eastern region than in the west.
quality and profit positively affect labor productivity, and technology gap does not, we fail to find significant evidence on the influence of capital intensity.An interaction of each of these variables with spillovers provides novel information.For example, fro m the interaction of HORI with T G we conclude that there is learning difficulty of do mestic firms fro m foreign firms.
Larger firms do benefit mo re fro m spillovers in industries where foreign firms have a larger presence.However, the study also reveals that profits contribute positively to the relationship between spillovers and the domestic labor productivity in the Ugandan manufacturing sector.The aforementioned findings reflect the need for well-designed macroeconomic and industrial policies to attract FDI inflows and to improve the c ompetitiveness of local firms.Since larger do mestic firms benefit more fro m spillovers in industries where foreign firms have a larger presence, domestic firms in Uganda should also be given the same incentives as those given to foreign firms to encourage fair co mpetit ion.Moreover, the finding of the locational advantage of the central reg ion in terms of lesser negative spillovers might be a reflect ion of the need for improved infrastructure and other investor-friendly environ ment in the east and western parts of Uganda to attract foreign investors to these regions.Finally, our results suggest that the joint ventures (foreign) should be pro moted if the manufacturing sector in Uganda is to reap not only fro m labor quality, firm size and pro fit spillovers, but also such an arrangement is likely to lead to less negative horizontal spillovers.Policies to steer collaboration between domestic and foreign firms in the manufacturing sector such as via trade fairs or advertising or patents would likely improve on the demonstration or imitation channel of horizontal spillovers.But it is also likely that foreign investors offer not only higher quality but also lo w-priced goods, leading to a shift of demand away fro m do mestic products and forcing local firms to produce at lower, less efficient capacity level, wh ich in itself makes competition translate into domestic productivity losses as it has a crowding -out or market-stealing effect (Ait ken and Harrison 1999; Pfeiffer et al., 2014).In this case, policies designed to engineer fair co mpetition and support an upgrade system of domestic firms still holed up in ancient inefficient technology due to institutional weaknesses deserve particular attention.Programs in line with skill acquisition through job train ing and the review of the curriculu m to focus on labor quality would be a strategy in the right direct ion.But there is also need, in our view, to change the biased mentality of do mestic firms that whatever is "Asian/African/Ugandan" in terms of technology is "inferior", as this is likely to deter learn ing and imitation for purposes of productivity improvement.Perhaps trade fairs, mass sensitization, advertisement and patent, that are inclusive of both domestic and foreign firms could be proper avenues via which better outcomes could be achieved.
While in the above analysis focuses on the drivers of labor productivity, several questions emerge in the course of investigation but need a separate study.For examp le, future micro studies would delve deeper into what in turn drives the contribution of foreign investors in the productivity -spillover analysis.Such an analysis is however likely to be limited by data.A co mparative analysis involving all the east African countries would in addition provide a better p icture on the catalysts of productivity; this is another area for future study.We also leave for further research an inter-industry analysis once data is available.

Table 2 .
Manufacturing sector in Uganda by ownership, location and origin o f foreign investors

Table 3 .
Descriptive statistics and pairwise correlat ion matrix

Table 4
Manufacturing sector in Uganda by subsector, origin o fFDI, location and ownership -2007-2010 Source: Co mpiled fro m Private Sector Investment Survey data (see data section)

Table 5 .
Drivers of labor productivityhorizontal spillovers and interactions

Table 6 .
Drivers of labor productivity -by sector and ownership