Are stricter investment rules contagious? Host country competition for foreign direct investment through international agreements

Abstract

We argue that competitive diffusion is a driver of the trend toward international investment agreements with stricter investment rules, namely defensive moves of developing countries concerned about foreign direct investment (FDI) diversion in favor of competing host countries. Accounting for spatial dependence in the formation of bilateral investment treaties and preferential trade agreements that contain investment provisions, we find that the increase in agreements with stricter provisions on investor-to-state dispute settlement and pre-establishment national treatment is a contagious process. Specifically, a developing country is more likely to sign an agreement with weak investment provisions if other developing countries that compete for FDI from the same developed country have previously signed agreements with similarly weak provisions. Conversely, contagion in agreements with strong provisions exclusively derives from agreements with strong provisions that other FDI-competing developing countries have previously signed with a specific developed source country of FDI.

Appendices

Appendix 1

See Table 10.

Table 10 List of source countries of FDI in sample and the number of treaties with hosts in 2002–2004

Appendix 2: List of host countries of FDI in sample

Albania, Algeria, Angola, Argentina, Azerbaijan, Bangladesh, Bolivia, Botswana, Brazil, Bulgaria, Burkina Faso, Cameroon, Chile, China, Colombia, Congo (Republic), Costa Rica, Croatia, Czech Republic, Côte d’Ivoire, Dominican Republic, Ecuador, Egypt, El Salvador, Equatorial Guinea, Estonia, Ethiopia, Gambia, Ghana, Guatemala, Guinea, Guyana, Haiti, Honduras, Hungary, India, Indonesia, Israel, Jordan, Kazakhstan, Kenya, Latvia, Lithuania, Madagascar, Malaysia, Mali, Mauritius, Mexico, Mongolia, Morocco, Mozambique, Namibia, Nicaragua, Niger, Nigeria, Oman, Pakistan, Papua New Guinea, Paraguay, Peru, Philippines, Poland, Romania, Russian Federation, Saudi Arabia, Senegal, Seychelles, Slovakia, Slovenia, Sri Lanka, South Korea, Sudan, Swaziland, Syrian Arab Republic, Tanzania, Thailand, Togo, Trinidad and Tobago, Tunisia, Turkey, Uganda, Ukraine, Uruguay, Venezuela, Vietnam, Zambia, Zimbabwe.

Appendix 3

See Table 11.

Table 11 Descriptive variable statistics

Appendix 4

See Table 12.

Table 12 Spatial lag variables

Appendix 5: FDI data issues

5.1 “Missing” FDI

The IMF’s Balance of Payments and International Investment Position Compilation Guide (IMF 2014), the OECD’s Benchmark Definition of Foreign Direct Investment (OECD 2008) and UNCTAD’s Training Manual on Statistics for FDI and the Operations of TNCs (UNCTAD 2009) all provide detailed instructions on how to compile FDI statistics and specify the involved complexities in voluminous manuals with hundreds of pages. Nevertheless, the reliability of actually available FDI data suffers from several shortcomings.

The instructions on international best practices have not only been revised repeatedly, but various developing countries can safely be assumed to lack the administrative expertise and capacity to follow the detailed instructions. As a matter of fact, even more developed countries differ in the degree to which they adhere to current best practices as given by the above mentioned institutions (IMF 2003).

All the same, we see good reasons to prefer bilateral FDI stocks as weights in the construction of spatial lag variables when assessing host-country competition for FDI through international agreements. First of all, the fact that both UNCTAD and the OECD report bilateral FDI statistics covering our large sample can be taken as the first indication that these institutions are confident to convey relevant, though inaccurate information by making these statistics available. Second, we tentatively assessed whether the magnitude of “missing FDI” is considerably larger than “missing trade,” which could be expected if FDI reporting was seriously deficient. Specifically, we compared the difference between (worldwide) outward FDI stocks declared by source countries and (worldwide) inward FDI stocks declared by host countries with the difference between worldwide exports and worldwide imports (details not shown here, but available on request). It is well known that world imports typically exceed world exports by about 2–5 %. The gap between inward and outward FDI stocks was much larger during the earlier part of our period of observation (about 20 % in the first half of the 1980s). However, the underreporting of outward FDI in these years is unlikely to affect our estimations since we refer to inward FDI stocks whenever available. Subsequently, the FDI gap continues to be larger than the trade gap (2–10 %, compared to 2–5 %); but “missing FDI” does not appear to be excessive compared to “missing trade”. See also Lipsey (2001: 27) who noted: “The worldwide discrepancy between outward and inward direct investment flows, which should be zero if all flows were recorded fully and consistently by both sides, has been no higher than 8 % in any year from 1993 to 1999, as contrasted with 40 or 50 % for portfolio investment.”

It is also noteworthy in this context that Barthel et al. (2010) find a fairly high correlation between inward and outward FDI stocks for those dyads in the FDI database which is also underlying our analysis with both the host and the source country reporting the relevant (inward and outward) FDI data (r = 0.86).

5.2 Valuing historical FDI stocks

As noted above, the adherence to best practices in reporting FDI stocks differs across countries and also has changed over time. Current IMF-OECD-UNCTAD guidelines recommend market values prevailing at the time of compilation as the basis for valuing FDI stocks. In practice, book values from the balance sheets of enterprises are often used as proxies of the market value of FDI stocks (for details, see IMF 2003: Sect. 7). If based on historical costs, such balance sheet values do not conform to the preferred market-price principle.

Importantly, using FDI stocks at historical costs does not necessarily imply that old FDI stocks would be weighted equally as most recently built FDI stocks in our calculation of spatial lags. For instance, the valuation of the direct investment position in US economic accounts, as specified by the US Bureau of Economic Analysis, explicitly accounts for accumulated depreciation to arrive at net property, plant and equipment at historical cost (for details, see: https://ideas.repec.org/p/bea/papers/0025.html). The IMF’s Balance of Payments and International Investment Position Compilation Guide (https://www.imf.org/external/pubs/ft/bop/2007/bop6comp.htm) lists Own Funds at Book Value (OFBV) as one of six methods for approximating market value for unlisted equity: “OFBV involves valuing a company at the value appearing in its books following International Accounting Standards (IAS). (…) IAS require most financial assets to be revalued on, at least, an annual basis and for plant and equipment to be depreciated” (IMF 2014: Appendix 4, paragraph A4.48).

It remains open to question, however, how many developing countries fail to meet demanding international standards on reporting FDI stocks. As noted by the IMF (2013: 123), “in cases in which none of the above methods are feasible, less suitable data may need to be used as data inputs. For example, cumulated flows or a previous balance sheet adjusted by subsequent flows may be the only sources available.” UNCTAD notes: “For a large number of economies, FDI stocks are estimates by either cumulating FDI flows over a period of time or adding flows to an FDI stock that has been obtained for a particular year from national official sources” (http://unctad.org/en/Pages/DIAE/Investment%20and%20Enterprise/FDI_Stocks.aspx). Hence, we perform an additional robustness test by subtracting from any year’s FDI stock the value of FDI stocks from 10 years prior (see main text for details).

5.3 Zero observations

In a dyadic setting, annual FDI data are frequently zero. What is more, it is often hard to decide whether zero observations are ‘real’ in the sense that partner countries explicitly report zero flows and/or stocks. In various instances, it cannot be ruled out that zero observations result from incomplete reporting (‘fake’ zeros).

It is in several ways that we reduce the risk of ‘fake’ zeros in our analysis. First of all, we use FDI stocks, rather than FDI flows, and we take averages over 3 years. The frequency of zeros, and thus the risk of ‘fake’ zeros, is considerably lower for FDI stocks than for FDI flows—particularly when considering period averages, rather than annual data. The risk of ‘fake’ zeros is further reduced in our analysis, compared to previous dyadic analyses. We consider developing countries only as hosts of FDI, not as sources of FDI. In other words, our analysis excludes dyads comprising two developing countries. Misreporting is most likely for these excluded dyads.

Furthermore, the preferred use of inward FDI in our analysis minimizes the risk that ‘fake’ zeros result in seriously biased spatial weights. Typically, missing entries in the reporting of FDI by host countries are less likely for more important sources of their inward FDI. Hence, it is unlikely that the two FDI terms in our formula for specific target contagion become zero because of incomplete reporting as long as the source country i is important enough for host country j and, respectively, competing host countries m. On the other hand, if one of the FDI terms in this formula is zero because of incomplete reporting, this does not cause serious bias as the missing source country i is most likely to be unimportant for host country j and/or competing host countries m. In other words, ‘fake’ zeros are unlikely to result in biased FDI-related weights unless decision makers have better information on the importance of specific source countries and competing host countries, compared to the researcher relying on official statistics.