Taxi, Takeoff and Landing: Behavioural Patterns of Capital Flows to Emerging Markets

ABSTRACT

This paper analyses volatility, persistence, predictability, correlation, comovement (or contagion risk) and sudden stop (reversibility) of capital flows (foreign direct investment (FDI), foreign portfolio equity investment, long-term and short-term debt flows) using time series econometric techniques for 24 emerging economies over 1970–2014. This is informative on the pattern and relationship between capital inflows, with implications for accommodating macroeconomic policies in countries receiving inflows. The paper also addresses the predictions of conventional theory, that differences are associated with the maturity of the capital (long-term vs. short-term), with the information-based trade-off model of Goldstein and Razin [(2006). An information-based trade off between foreign direct investment and foreign portfolio investment. Journal of International Economics, 70(1), 271–295], that differences are associated with the structure of the capital (equity vs. debt). In line with the latter, equity flows (FDI and portfolio) are less volatile and persistent, more predictable and less susceptible to sudden stops than debt flows. Contrary to conventional theory, short-term flows are not more volatile, but there is evidence that correlations and risks of contagion are strong within all capital flow components.

KEYWORDS:

• Capital flows
• volatility
• persistence
• forecasting
• comovement
• sudden stop

JEL CLASSIFICATION:

• F21
• F32
• F34
• F37
• G01
• O57

Disclosure Statement

No potential conflict of interest was reported by the author.

Notes on contributor

Bilal Keskinsoy is a lecturer in economics at Anadolu University. Holds a PhD in economics and an MSc in economics and econometrics from the University of Nottingham. Earned BAs in business administration and economics from Yeditepe University.

Notes

1. Not just in case of emerging and developing countries but also for advanced nations capital flows are, at times, regarded as troublesome. The following excerpt (from a speech of the former US president, Bill Clinton) in Karolyi (2004) exemplifies this: ‘ …  it is now time for the world to take the next steps of implementing a new financial architecture and long-term reform of the global financial system. This should include steps to reduce the entire financial system's vulnerability to rapid capital flows … ’

2. See McKinnon and Pill (1998), Kindleberger and Aliber (2005).

3. There are also policy prescriptions to contain and control volatility and speculative activity in international financial markets and capital flows. In his presidential address at the conference of the Eastern Economic Association in 1978, Washington DC; James Tobin proposed a tax (credited to him as ‘Tobin tax’) on capital market transactions to mitigate excess fluctuations and stampedes across the markets and thus to serve as ‘sand in the wheels’ of international finance. Tobin (1978) and Eichengreen, Tobin, and Wyplosz (1995) formalize the idea, while Uppal (2011) reviews its costs and benefits. José Manuel Barroso, then European Commission president, has called for a tax for similar purposes on financial transactions throughout the European Union or at least for the Euro Zone (Financial Times, 28 September 2011).

4. The view that FPEI is reckoned as a kind of short-term capital flow could be traced in Stulz (1999) who states that in a positive feedback trading prevalent stock market environment highly liquid short-term financial instruments are open to volatility spawning speculative trading. Also see Sachs et al. (1996) as well as ‘hot money’ on Wikipedia.

5. Information asymmetries in the model are envisaged to take place at two stages. At first stage there is a principal-agent kind of information asymmetry that exists between entrepreneurs and managers where the level of ownership reduces the costs and improves efficiency by mitigating the effects of information failures. The second stage information asymmetry arises between the current owner and the potential buyer when the former happens to sell prior to the maturity. The investor liquidity needs and preferences, instead of market conditions, determine the ‘liquidity shocks’ definition of the paper.

6. Under the assumptions of financial frictions and partial inalienability, Albuquerque (2003) offers an FDI–non-FDI dichotomy for capital flow volatility profiles.

7. For instance, Liu et al. (1999) quantify volatility by logarithmic growth rates of the series.

8. Cox and Sadiraj (2010) argue that the coefficient of variation has poor normative and descriptive performance in risk appraisal and Polly (1998) shows that there is size-related bias in the coefficient of variation. See also Sørensen (2002).

9. Volatility of the cyclical component obtained from the partitioning through HP filter could also be seen analogous to the random component in Chari and Kehoe (2003) and stochastic component in Acemoglu and Zilibotti (1997).

10. Boyson, Stahel, and Stulz (2010), Pukthuanthong and Roll (2009), Bekaert, Hodrick, and Zhang (2009), Bordo and Murshid (2006) and Mauro, Sussman, and Yafeh (2002) also employ principal component analysis to gauge comovement and contagion.

11. As stated in Wooldridge (2002), achieving a -consistent estimator of β is possible by maximizing the partial (sometimes quasi, pseudo or only probit) log-likelihood function .

12. I opt to use the net liability balances as they constitute the data of interest. The idea is that, as Dornbusch (2002) argues, when foreign financing to a particular sector is withdrawn (i.e. a sudden stop or reversal occurs) that means a capital outflow and not a substitution into other assets currently held by that sector.

13. Binici, Hutchison, and Schindler (2010), Aykut, Francis, and Tereanu (2010), Neumann, Penl, and Tanku (2009), Alfaro, Kalemli-Ozcan, and Volosovych (2008), Schularick (2006), UNCTAD (2000) adopt similar practice.

14. As an alternative presentation of the raw data two-way line plots of the time series are illustrated in the appendix.

15. On the vulnerability generating impact of short-term external borrowing see Kose et al. (2009), Dornbusch (2002), Chang and Velasco (2001), Cole and Kehoe (2000), Calvo (1998).

16. See Martin and Rey (2006), Caballero and Krishnamurthy (2006), Aizenman and Pinto (2005), Aghion et al. (2004), Chari and Kehoe (2003), Sarno and Taylor (1999a), McKinnon and Pill (1998).

17. The surmise has been developed and discussed around the concept of ‘debt intolerance’ in Reinhart, Rogoff, and Savastano (2003), Reinhart and Rogoff (2004), Eichengreen, Hausmann, and Panizza (2007).

18. Three alternative unit root tests, (Dickey & Fuller, 1979; Phillips & Perron, 1988; Kwiatkowski et al., 1992), are employed to determine the order of integration of a series. The decision is made for a particular outcome (i.e. the order of integration) if the results from at least two tests are statistically equivalent. This is consistent with the account of Stock (1994) that interpretation of the unit root tests is a matter of judgement. Results from unit root tests and correlograms with Q-statistics are in the appendix.

19. Similar observation qualitatively noted for portfolio equity flows in Stulz (1999).

20. Similar and alternative definitions could be found in Bordo, Cavallo, and Meissner (2010), Cavallo and Frankel (2008), Honig (2008), Hutchison and Noy (2006).

21. Of the similar nature, frequency of sign changes in capital flows are given in Table 6.4 in Lipsey (1999) and the number of sudden stops presented in the first tables in Honig (2008), Hutchison and Noy (2006) and in Figure 1 of Cavallo and Frankel (2008).

22. Shleifer and Vishny (1992), Manso (2008), Moyen (2007), Aivazian, Ge, and Qiu (2005), Hennessy (2004).

23. Although investment effect of the country level debt overhang is documented to be ambiguous, the actual service of the debt is found to crowd out investment. See Bulow and Rogoff (1990), Warner (1992), in particular, Cohen (1993).

24. See Fisher (1933), McKinnon and Pill (1998), Schneider and Tornell (2004) and Mendoza (2010).

25. For identical recommendations please refer to Eichengreen (2000) and Cardoso and Dornbusch (1989).

Additional information

Funding

This work was supported by the Ministry of National Education (MEB) of the Republic of Turkey (TC) [Grant Number 3509].