Portal de Periódicos da CAPES

The Effects of Political Risk on Different Entry Modes of Foreign Direct Investment

2014; Taylor & Francis; Volume: 40; Issue: 5 Linguagem: Inglês

10.1080/03050629.2014.899225

1547-7444

Hoon Lee, Glen Biglaiser, Joseph L. Staats,

Corporate Finance and Governance

AbstractAlthough numerous studies document the effect of political institutions on foreign direct investment (FDI), few works in the political economy literature have investigated the link between political institutions and the mode of entry chosen by investors, be it mergers and acquisitions, joint ventures, or greenfield investments. Using panel data for 111 developing countries covering 1980–2006, we find that countries with political institutions that uphold good governance tend to attract higher levels of mergers and acquisitions, as opposed to joint ventures and greenfield investments, because such institutions help to mitigate the special risks faced by merger and acquisition investors. Our findings provide a nuance for understanding the different effects of political institutions based on the particular mode of entry.KEYWORDS: foreign direct investmentgreenfield investmentsjoint venturespolitical risk Notes1 See also Sauvant (Citation2012), who discusses how more host countries consider specific M&As as less beneficial than GIs.2 Globerman and Shapiro (Citation2005) looked at whether the determinants of M&As as the mode of entry were the same as for FDI overall. They found that for the most part they were, although the coefficients differed, indicating varying importance of determinants depending on mode of entry. However, they only looked at M&As in relation to FDI. They also only utilized pooled data over a single time span rather than constructing models derived from panel data over multiple years. Henisz (Citation2000) also considered mode of entry but in the context of potential expropriation hazards to MNCs from operating in joint-venture partnerships where foreign firms either held minority or majority equity stakes.3 Our study addresses the mode of entry literature on the basis of directly owned production, but there are other types of entry, including arm's-length production (for example, engaging in licensing or subcontracting) that for lack of data we are unable to investigate (see Williamson Citation1985).4 See Oxley (Citation1999:307), who notes how infrequently alliances are formed between firms in developed and developing countries.5 See also Oxley and Yeung (Citation2001), who argue that MNCs located in countries with weak rule of law may wish to ally with other firms.6 See also Anderson and Gatignon (Citation1986), Delios and Henisz (Citation2000), and Hennart (Citation1988), who find that firms can reduce the level of host-country risk by using lower ownership modes.7 Exuberance in search of jobs, even when political risk is not a concern, is illustrated by Germany paying Dow Chemical $3,400,000 for each new employee (Morisset Citation2003:3).8 There are also likely instances where firms have operated in a country before or have current operations there, which thus affects their mode of entry decision. Based on our data, we cannot test the effect of prior influence, but our expectation is that, in general, foreign firms are more likely to choose M&As as a mode of entry in countries with better quality of governance.9 In the case of Japan, Walmart initially acquired a 6.1% stake in Seiyu, but it became a wholly owned subsidiary a short time later.10 The 12 Sub-Saharan countries are Botswana, Ghana, Lesotho, Malawi, Mozambique, Namibia, Nigeria, South Africa, Swaziland, Tanzania, Uganda, and Zambia. In Central America and Sub-Saharan Africa, Walmart acquired existing firms, but it took small stakes, obtaining 33% control in Central America and a majority position in Sub-Saharan Africa, not gaining 100% shares as it did in the M&A countries. Walmart operated with other actors in making decisions, and thus Central America and Sub-Saharan Africa are examples of JVs.11 Although India appears to be an exception, as it possesses higher respect for the rule of law as compared to the other JV cases, it is also one of the few countries in the world that imposes restrictions on M&As, which likely explains why Walmart entered India using a JV strategy.12 Some may ask if the potential pool of domestic firms that are viable targets affects M&A decisions. We argue that even with a weak pool of domestic firms, some M&As would still occur, if for no other reason than to obtain information about the country's consumers, gain access to supply and distribution chains, or even find opportunities to earn monopoly rents in the host country. The case of Walmart is illustrative, as the firm could have chosen an M&A strategy that resulted in it holding 100% stakes in some of the poorest countries in the world. Instead, Walmart opted for JVs in poorer countries, including El Salvador, Guatemala, Honduras, Lesotho, Malawi, Mozambique, Nicaragua, Swaziland, and Tanzania, as well as countries that are not so poor (Brazil, India, and Mexico). Some may also wonder if investment restrictions are more common in authoritarian systems and that the link between democratic institutions and M&As is driven more by national governments' earlier policy choices than by the MNCs' responses to policy risk. We consulted UNCTAD (Citation2009) and found that out of more than 120 countries, only the following 16 imposed restrictions on mode of entry: Algeria, Australia, Bahamas, Brazil, Canada, China, India, Indonesia, Japan, Jamaica, Laos, Malaysia, Nigeria, the Philippines, Sao Tome and Principe, and Venezuela. Only a tiny percentage of countries have mode of entry restrictions, and based on the list, regime type and poverty issues, appear to play little role.13 The last year of data available for our main dependent variables is 2006. The data also are not refined enough to adjust for the firm's country-of-origin, which may influence mode of entry.14 Although M&A data are collected by UNCTAD, the data originally come from Thomson Financial. Also, the M&A data measure the value of the transaction at the time it is finalized, which is therefore positive. In order to generate the relevant ratios, we follow Globerman and Shapiro (Citation2005:84–85) and make two adjustments. First, when FDI flows are negative values (due to net inflows), but M&A are recorded as positive, we recorded the M&A/FDI ratio as equal to one (the results remain similar after dropping these cases). Second, when M&A amounts for a country in a given year exceed total FDI flows, we reassigned a value of one. Otherwise, the results lead to implausibly large ratios due to either very small FDI flows or very large M&A amounts. We also adjusted our data for JVGI/GDP. When JVGI (subtracted from total FDI flows) takes a negative value, we recorded it as equal to zero for similar reasons offered in M&A/FDI ratio.15 See Conybeare and Kim (Citation2010a; Citation2010b) for details about the data.16 Data for announced and completed GI deals are unavailable, which explains why we only compare M&As and JVs here.17 Because data are available every other year until 2002, we interpolate 1997, 1999, and 2001.18 See Büthe and Milner (Citation2008) for insights on the role BITs play is supporting investor safety.19 We also wanted to include government restrictions on foreign investments as a control variable, but the available data either do not differentiate by mode of entry and cover less than half of the developing countries that we study (Pandya, Citationforthcoming) or do not have entry data prior to 2009 (UNCTAD Citation2009). In addition, the UNCTAD data indicate very few developing countries placing any restrictions at all on entry of foreign investment.20 For obvious correlation reasons, we exclude Democracy from models where democratic stability is the independent variable of interest.21 The summary statistics for the variables are available in the online appendix.22 For 3 of the 18 models, we use a panel-specific AR(1) parameter because the models would not run using the common command. Replacing panel-specific for common in the other 15 models yields nearly the same results, suggesting that the AR(1) structure makes little difference for our overall findings.23 The results are available from the authors.24 We exclude GDP as a control variable in models where GDP is a denominator in the dependent variable for possible collinearity reasons.25 We also ran Tobit with respect to the ratio of completed over announced or over total deals, and we obtain similar results to those found using PCSE. For space reasons we do not report the results, but they are available upon request.