Does Foreign Ownership Increase Firms’ Productivity? Evidence from Firms Listed on Amman Stock Exchange

1 Introduction

Jordan, as many other Arab countries, adopted structural adjustment programmes (SAPs) under the guidance of the World Bank with the expectations that liberalizing, restructuring and opening the Jordanian economy will bring enormous gains, including updating and improving the country’s industrial base and technologies.^[1] The implementation of the SAPs accelerated in the late 1990s and prompted the adoption of friendly foreign-ownership legislative environment and the privatization of some of the largest state-owned enterprises.^[2] As a result, foreign ownership of firms listed on the Amman Stock Exchange (ASE) has increased substantially. This paper attempts to evaluate the theoretically anticipated gains of foreign ownership in terms of improving the efficiency of firms’ productivity. However, using a production function framework, this study fails to find evidence on improvements of the production efficiency for firms with foreign ownership. Multinational enterprises (MNEs) acquired well-established Jordanian companies, and hence, once the estimation method applied in this study controls for the unobservable firm fixed effects and selection bias, productivity gains from foreign ownership becomes statistically insignificant. This result indicates that multinationals acquired Jordanian firms to exploit these companies’ advantages rather than their own technological advantages which resulted in little or no knowledge transfer from the multinationals to their Jordanian affiliates.

Foreign investment can have an important impact on domestic firms’ productivity growth and technologies. Multinationals arguably possess a technological advantage (Hymer 1976) and hence their presences in a host market can contribute to knowledge transfer and diffusion in that market directly or through spillover effects. MNEs directly transfer knowledge to their affiliates by providing them with advanced machinery and equipment, by implementing updated production processes and by introducing new products (Konings 2001). In addition, implementing new technologies requires labour training which in turn improves labour skills and enhances efficiency (Greenaway, Upward, and Wright 2002). Also, MNEs bring new and efficient styles of management, marketing, governance and accounting (Fu 2008; Konings 2001). Hence, domestic firms can acquire new knowledge, and consequently improve their productivity, by altering their ownership structures to include foreign investors.

However, the extent and nature of benefits arising from the transfer of knowledge from MNEs to their affiliates depend on a number of important factors, including the host economy’s local conditions and the quality and motivation of the foreign investment. The extant literature suggests that positive externalities in a host market created by foreign ownership are conditioned on threshold effects such as domestic level of human capital and overall quality of the market’s institutions (Girma 2005; Alfaro, Kalemli‐Ozcan, and Sayek 2009).^[3] One relevant aspect of institutional quality to this study is the degree of financial development. Jordan, however, ranks among the top of Arab countries in terms of bank development indicators with good scores in private credit to GDP and size of bank assets (Omran and Bolbol 2003; Krogstrup and Matar 2005). In addition, the ASE is considered a well-developed market compared to other markets in the Middle East and North Africa (MENA) region (Krogstrup and Matar 2005). For example, Lagoarde-Segot (2009) reports that the ASE displays low market volatility and low market illiquidity in comparison to 28 emerging markets.

Another area of research investigates the confounding influences of the quality and motivation of foreign investment on the anticipated productivity gains. For example, technology transfer is viable when a multinational firm sets up an affiliate to take advantage of its unique firm-specific assets while at the same time protecting its propriety knowledge through its ownership of the affiliate (Dunning 1993; Gorg and Greenaway 2004). However, the multinational presence is not expected to induce technology transfer if the foreign investment is driven by the multinational’s incentive to have access to new technology in the host market (Driffield and Love 2007). In addition, an observed positive association between foreign presence and high productivity can be explained by the MNE acquiring strategy under which it targets well-performing domestic firms (Evenett and Voicu 2001; Arnold and Javorcik 2009; Karpaty 2007). This paper fits the most into this line of research as it examines the impact of foreign ownership on the productivity of Jordanian firms, with a focus on dealing with the unobserved firm heterogeneity and self-selection bias. MNEs are likely to target Jordanian companies with unique advantages such as control over natural resources, large market shares and access to local and regional markets. These advantages are also likely to be correlated with high production efficiency. If MNEs acquired well-performing Jordanian firms to exploit these companies’ advantages, then once the firm’s fixed effects and selection bias are controlled for, the positive impact of foreign ownership on productivity becomes insignificant.

Examining the gains of productivity in the Arab region is interesting because Arab countries have been performing poorly in attracting FDI inflows (Sadik and Bolbol 2001; Omran and Bolbol 2003; Krogstrup and Matar 2005). Table 1 reports FDI inflows to developing countries grouped by region.^[4] This table shows that FDI inflows have been increasing steadily in all regions in the past two decades except for the years around the financial crisis. However, the size of FDI inflows varies greatly across the different regions of the world. The MENA region in particular has received the lowest amount of FDI inflows compared to developing countries in all regions in almost all years in the period 1995–2011.^[5] Also, the size of FDI inflows received by the MENA region in 2011, the start of the Arab spring, dropped to nearly half the average of the previous five years. Although it is early to conclude that this decrease in FDI inflows will persist, the political unrest in the region suggests that the drop in FDI inflows may continue for the next few years.

Some observers believe that countries in the MENA region are missing out on the anticipated gains of foreign ownership in terms of acquiring new technologies and on improving their firms’ production efficiency (Sadik and Bolbol 2001; Omran and Bolbol 2003; Krogstrup and Matar 2005). This is especially relevant given that the empirical evidence from an ample number of developing and transition economies indicates that there is a positive impact of foreign participation on a firm’s productivity (Aitken and Harrison 1999; Chudnovskya, Lópeza, and Rossia 2008; Damijan et al. 2003; Smarzynska-Javorcik 2004; Fu 2008; Liu and Wang 2003). This evidence is fairly robust to the choice of economic regime, time period, and method of estimation.^[6]

In order to encourage FDI many Arab countries, including Jordan, have adopted investment incentive programmes (IIPs) that offer tax holidays and/or tax and tariff reductions.^[7] However, these investment incentives are costly and are economically justifiable only when the gains of foreign investments exceed the costs of investment incentives. For example, tax holidays and tax reductions result in a substantial decrease in tax revenues (Clark, Cebreiro, and Bohmer 2007). If that decrease in tax revenue is not offset by gains from new capital investments, induced by the incentive programme, the government may cut its expenditures or raise taxes elsewhere. This in turn can have an immediate negative impact on the economic conditions of the public, especially low income households (Chen 2004). In addition, the incentive programme may distort the incentives of non-recipient firms and hence allocate the country’s resources inefficiently (Clark, Cebreiro, and Bohmer 2007). Also, investment incentives can redistribute income from host countries to MNEs especially when countries are competing for FDI (Gorg and Greenaway 2004). Therefore, productivity gains of foreign ownership must be large enough to justify these costly incentives. The analysis presented in this paper evaluates empirically the productivity gains resulting from foreign ownership by employing a productivity function with a technological innovation term using a sample of manufacturing and mining companies listed on the ASE. The empirical findings indicate that the productivity of companies with foreign ownership is not significantly different from their domestic counterparts implying that Jordanian firms are not gaining from foreign ownership. Hence, the analysis presented in this paper questions the financial viability of Jordan’s investment incentive programme designed to encourage foreign direct investments (FDIs).

In this paper I focus on the case of Jordan because of various reasons. First, Jordan has been receiving large amounts of FDI inflows as a percentage of its GDP compared to other Arab countries as indicated by figures presented in Table 2.

In addition, FDI inflows to Jordan translate into a large contribution of foreign ownership in the number of employment, sales, market values and subscribed shares in listed Jordanian companies. The first four columns in Table 3 indicate that the contribution of firms with foreign ownership to total employment is 61%, to sales is 76%, to market value is 83% and to subscribed shares is 64%. In the next four columns I compute more conservative contributions based on the equity holding of firms with foreign ownership. However, firms with foreign ownership still account for a large contribution of employment, sales, market value and subscribed shares of 25%, 27%, 47% and 30%, respectively.

Second, the results and conclusions of this study are useful in drawing policy implications not only for Jordan but also for other Arab countries. This is because Jordan shares with Arab countries similar market characteristics. The manufacturing sectors in Arab countries are labour-intensive and consumer-oriented with their exports comprising mainly of natural resources. The final motivation for the use of Jordan relates to data availability. Foreign ownership data is publicly available for Jordanian listed companies and hence it allows the study of the impact of FDI on firm’s productivity using firm-level data. The use of firm-level data is advantageous since the aggregation of data over economies results in a substantial loss of information (Damijan et al. 2003).

2 Research Design

3 Data and Descriptive Statistics

I use firm-level, hand-collected data from the Corporate Guides published by the ASE and from companies’ financial statements. The data covers the period 2001–2011 for an unbalanced panel of Jordanian firms listed on the ASE in the manufacturing, utilities and construction sectors. Private companies (domestic or with foreign ownership) are not required by law to disclose financial information and hence this study is limited to companies with listed shares in the ASE. In addition, this study focuses on manufacturing, utilities and construction companies because companies in these sectors are the ones expected to benefit the most from the technology transfer through acquiring modern equipment, machinery, production processes and employee training to replace their obsolete and outdated technologies (Konings 2001). Finally, the study eliminates companies with a number of employees less than 10 (Damijan et al. 2003).

The characteristics of the sample firms are presented in Table 4. The average foreign ownership is 18.07%, which indicates that foreign investors own on average 18.07% of the sample firms’ equity capital. The average foreign ownership is higher than the literature’s arbitrary cut-off of 10% which is assumed to represent a foreign controlling interest in the firm. At the 10% cut-off the average ownership of foreign investors increases to 35.4% of the firm’s equity capital. These figures confirm the earlier observation that foreign participation in Jordanian firms is large. The average deflated output produced by the sample firms during the study period is only 37.5 million JDs. In addition, half of the firm-year observations have less than 5.31 million JDs in sales. Table 4 also reports that the average number of employees, fixed assets, market value and total assets are all moderate. However, these numbers are not surprising given that Jordan’s economy is relatively small with, for example, a GDP in 2011 of $28.84 billion (the World Bank). In terms of labour productivity, the average and median labour productivity in JDs (not reported) are JDs 72,027 and JDs 33,671 per worker.

To examine if there is a significant difference in the output y, labour productivity (LPROD) and value added per employee (ValueAdded) between firms with foreign ownership and domestic firms I run tests on the difference in means and the Wilcoxon test. The results are reported in Table 5. Panel I reports the results of the difference in means. The results indicate that firms with foreign participation have significantly larger output y and ValueAdded. However, the labour productivity LPROD is not significantly different between firms with foreign ownership and domestic firms. Table 5, Panel II reports the Wilcoxon test. Similar to the previous results, the Wilcoxon test is significant for y and ValueAdded but not LPROD.

Finally, the correlation matrix is reported in Table 6 and indicates that y, LPROD and ValueAdded are all significantly correlated with foreign ownership.

4 Results and Analysis

To test the impact of foreign ownership on a firm’s productivity I start by estimating eq. [5]. Table 7 reports the F statistic which tests the joint restriction that v_i are equal. The F statistic for the estimations using y and LPROD, respectively, is significant at the 1% level. Therefore, the null of no individual heterogeneity is rejected and I proceed to use panel data estimators. The fixed effects estimator treats v_i as fixed constants while in random effects estimator v_i are assumed to be drawn randomly. In this study, I do not make assumptions about the correlation between the independent variables and the unobservable effects. Hence, I estimate eq. [5] with fixed and random effects respectively and run the Hausman test and report the test results in Table 7. The Hausman test for the estimations using y is significant at the 1% level and is insignificant LPROD. Therefore, I report the results of fixed and random effects.

Table 7, Panel I, reports the results of the impact of foreign participation on the firm’s productivity using fixed and random effects and Tobit regression.^[9] The standard errors are robust and clustered by industry. The coefficient of the variable FOWN is positive with a small magnitude and statistically insignificant across all specification. This result suggests that firms with foreign ownership are not more productive compared to domestic firms. In another set of results (not reported) I account for the possibility that technology may be transferred through efficiently utilizing labour or capital. I account for this possibility by additionally including interaction terms between foreign ownership and capital FOWN_it–1*k_it and foreign ownership and labour FOWN_it–1*l_it into the vector of explanatory variables. The coefficients on these interaction terms are positive but statistically insignificant. This result indicates that foreign ownership does not improve the efficiency of utilizing labour or capital. In relation to the other variables, the results are consistent with the international evidence. The coefficients of the variables k and l have the expected positive sign and are significant (though to various degrees). These results indicate that capital k and labour l contribute significantly to the output y.

Panel II of Table 7 reports the results of estimations similar to the ones reported in Panel I but using a subsample of 210 observations. In this subsample I include in a new variable Skilled that captures the human capital quality (Zhao and Zhang 2010). This variable is measured as the logarithm of the ratio of skilled to total labour (Chudnovskya, Lópeza, and Rossia 2008). I collect the data from the companies’ annual reports for the period 2008–2011.^[10] The purpose of including this variable is to take into account other factors that influence productivity. The main concern is to examine if the coefficients of the base-model variables are stable, especially the main variable of interest, FOWN. The results reported in Panel II indicate that the coefficient of the variable Skilled is insignificant when using y; significant at the 10% level only in random effects when using LPROD; and significant at the 1% level but with a coefficient size close to zero when using ValueAdded. The coefficient of the variable FOWN reversed its sign to negative but still has a small magnitude and is statistically insignificant in all models. The other two variables, l and k, have their expected signs and are significant in all models (except for k in the third column). The mixed results on Skilled and the stability of other factors allow us to proceed using the base model defined in eq. [4].^[11]

As discussed in Section 3.3, eq. [5] may be subject to a selection bias. To account for this bias I estimate eq. [6]. In order to estimate eq. [6], I create an indicator variable, FOWN_D, that takes the value of 1 if the foreign participation is above 10% and 0 otherwise. Hence, the treated group contains firms with foreign ownership and the control group contains domestic firms. The results are reported in Table 8. The reported Lambda is statistically significant which indicates that there is a selection bias, that MNEs ownership decision is not random. The decrease in the number of observation is due to missing data on the Companies Guide on some variables in the selection equation.

The main variable of interest, FOWN_D, is positive but statistically insignificant. I obtain this result using y, LPROD and ValueAdded, respectively, as the dependent variable. This result confirms the previous one reported in Table 7 that firms with foreign ownership are not more productive compared to domestic firms. As for the other determinants of output y and labour productivity LPROD, I report similar findings to the ones reported in Table 7. The coefficients of the variables k and l are significant with the expected positive sign (except for k in the third column). In addition, Table 8 reports the results of the selection model (the determinants of foreign ownership). It is useful to look at the significance and the signs of the variables determining MNEs decision to participate in domestic firms.^[12] The findings indicate that MNEs acquire the equity of domestic firms which are large (positive and significant Size); with low payout ratio (negative and significant DY); and with undervalued assets (negative and significant MB). These results are consistent with the evidence reported by Dahlquist and Robertsson (2001) and the evidence from the Jordanian market reported by Abu-Ghunmi et al. (2013). These findings support the hypothesis that foreign investors “cherry-pick” their target companies by acquiring reputable, value firms with assets that the multinational wants to control.

As a further test of the impact of foreign ownership on the firm’s productivity I use PSM for pairing observations. Table 9 reports the average treatment effect for the treated, ATT. The ATT is insignificant using y, LPROD and ValueAdded. This finding confirms the previous results that once the unobserved heterogeneity and selection bias are controlled for, the production efficiency of firms with foreign ownership and domestic firms are not significantly different.

Finally, I account for the possible endogeneity between output and inputs in the production function using system-GMM estimation method instrumented with three lags. I report the results in Table 10. The variable of interest FOWN is still of small magnitude and statistically insignificant.

5 Conclusion

Arab countries attract only a small fraction of the total FDI flows to developing countries. Jordan, however, have received substantial FDI flows relative to its GDP and those flows translate into a large contribution of foreign investors in the number of employment, sales, market values and equity capital in listed Jordanian companies. Based on the premise that foreign investors can transfer advanced technologies and knowledge to their affiliates in the host market at a low cost one would expect that firms acquired by foreign investors will outperform their domestic counterparts in terms of production efficiency. Using TFP, labour productivity and value added per employee as measures of productivity innovation and efficiency in a sample of Jordanian companies, I fail to find evidence on improvements of production efficiency for firms with foreign ownership. The finding of no-impact is robust using alternative estimation methods. This result contradicts the documented international evidence from other developing countries. However, this finding can be explained by the motivation of ownership of foreign investors who are targeting well-performing companies with valuable assets to be acquired under their control. The no-result finding of this study questions the value of current costly investment incentives designed to attract FDI. In addition, it points out to the complexity of realizing possible gains of foreign ownership especially in the context of a small economy.