3. Main macroeconomic theories and empirical evidence
3.3. Capital controls and structural determinants
Before proceeding with a review of what the capital controls debate brings in evidence of their existence, it is useful to put it in context with the strands of literature that have been reviewed so far. Von Furstenberg (1998) splits definitions of international financial integration into two categories: those relating to
“prerequisites for” and those based on “consequences of” such integration. So far, only the second class is presented, which deals with testable consequences of financial integration deduced from atemporal rate arbitrage and intertemporal optimization models. Indeed, the capital controls discussion belongs to the first class, for this approaches the subject of integration from the perspective of necessary preconditions. It implies that international financial integration would be complete were it not for the existence of capital controls and other barriers.
The debate on capital controls goes back longer than attempts to measure international financial integration. The Tobin tax is for example already proposed in 1978. The desire to measure the degree, speed and process of capital mobility in the 1980s and 1990s is born out of the discussion on capital controls. All the while though, capital controls continue to be vigorously debated alongside, but where this debate takes on a distinctly different character. The capital controls discussion focuses primarily on qualitative issues: the desirability and effectiveness of capital controls; the form that these capital controls should take; their feasibility and likely effects on the economy; special macroeconomic and political circumstances in which capital controls may be justified; and how capital controls are positively reconciled in a general-equilibrium framework.3 For the purpose of this research, to obtain evidence on the extent of the integration of financial markets, I leave this qualitative part of the capital controls debate aside and rather focus on the evidence this strand has produced on the existence of capital controls and any measures of quantifying this.
Evidence of the existence of capital controls is generally quantified through the construction of indexes for a panel of countries with the help of IMF annual reports on exchange arrangements and exchange restrictions. The IMF publishes this report since 1950 and for a great many countries reports whether restrictions on capital account transactions are in place, special foreign exchange rates apply for some or all capital transactions or invisibles. Epstein and Schor (1992) are among the first to calculate such an index for 16 OECD countries over the period 1967–1986. The index represents the number of restrictive
3 For a flavor of this qualitative debate on capital controls, see Eichengreen et al. (1995), Garber and Taylor (1995), Edwards (1999) and Epstein and Schor (1992), wherein each of these arguments features.
practices in place in each country in each year with a maximum of two. Epstein and Schor (1992) observe that the motivation for capital controls seems to be closely linked to political and institutional factors but leave this notion otherwise unexplored. This is instead further elaborated on in Alesina et al. (1994) who empirically investigate the relation between the presence and the removal of capital controls with the structural economic and political features of countries. This type of research has become known in the literature as the structural determinants of capital controls. Alesina et al. (1994) use a sample of 20 OECD countries for the years 1950 to 1989 and with the same IMF data construct a capital controls dummy (which takes the value of one when capital controls are in place and zero otherwise). They then perform a (probit model) regression of this capital controls dummy on a set of possible explanatory variables that include dummy variables for when a fixed exchange rate regime is in place and others to indicate government strength and central bank independence. The evidence reveals an a priori expected result, which is nevertheless interesting: that capital controls are more likely imposed by a strong government and when the central bank is less independent. The latter relates to the fact that governments benefit from an inflation-tax with higher seigniorage revenues from imposing capital controls. In these circumstances capital controls are able to keep real interest rates artificially low. This is linked with the deviations from interest parity discussion and this should then show up in such calculations but Alesina et al. (1994) unfortunately do not perform this test. They do find instead that capital controls are more likely at times when a fixed exchange rate regime is in place and in countries where the agricultural sector is bigger for which a variable is also included.
Milesi-Ferretti (1998), part of the set of authors of the original article, then extends the Alesina et al. (1994) analysis to a broader panel data set of 61 industrial and developing countries for the period 1966 to 1989. Further extensions of his study are threefold. First, splitting the capital control dummy into two dummy variables, one for each type of restriction the IMF provides data for in its annual reports. Secondly, including a whole range of novel explanatory variables. Thirdly, using a slightly modified estimation model (a logit as well as a probit regression model) and in addition to the annual series also five year non- overlapping averages to reduce serial correlation and to smooth out temporary shocks. Milesi-Ferretti’s results confirm the Alesina et al. results in many ways and particularly that capital controls are more likely to be in place when monetary policy is more firmly under the government’s control. Additional findings are that poorer countries and countries with larger governments are more likely to adopt capital controls. Given that only the IMF data on the use of capital controls is on offer for a wide range of countries and over a relatively long period of time, the indexes are probably the best that can be produced in terms of a measure of the extent of capital controls. They are comparable and easy to interpret. But the
limitations of the IMF data also translate to the limitations of the indexes. Edwards (1999) notes in this respect that the indexes are very general, do not measure the intensity of capital restrictions, fail to
distinguish between the type of flows that is being restricted and ignore the fact that legal restrictions are easily circumvented. Epstein and Schor (1992) further note that a drawback of the IMF definition is that it does not include some practices of countries which might reasonably be considered capital controls. A possible alternative approach is shown by Lemmen and Eijffinger (1996) who measure the intensity of capital controls with closed interest parity deviations. Here any positive (negative) deviations are
associated with capital import (export) restrictions. The drawback of this approach is that it can only make the distinction between these two broad types of capital control but cannot be more specific than that. Furthermore, it is assumed that all interest deviations are due to capital controls. An alternative route is for any researcher of capital controls to compile her own data on capital controls that have been employed in countries with a more detailed categorization of capital controls rather than rely on the limited IMF data. This is very costly and time-consuming indeed.