The influence of interest: Real US interest rates and bilateral investment treaties

Forbes 3/5/2014. Why The Worst Is Still Ahead For Turkey’s Bubble Economy.

http://​www.​forbes.​com/​sites/​jessecolombo/​2014/​03/​05/​why-the-worst-is-still-ahead-for-turkeys-bubble-economy/​

Die Zeit, January 2014.

As in all of the following, we exclude OECD countries from the sample as a means of limiting our analysis to primarily capital importing countries. (Another rationale for excluding high-income countries, with more stable and mature economies, is that they should be less affected by changes in US monetary policies; see also Di Giovanni and Shambaugh 2008.) Except for the initial OECD members, a country is included in the sample up to the point when it joined the OECD. The equation of OECD membership with “capital exporting” is, of course, inexact. And while this coding rule benefits from conceptual parsimony, some countries – Turkey for most of the 20th century, for example – are likely mislabeled by it. In practice, this rule does not seem to matter much to the reported regressions: We did not find substantially different results when using alternative proxies for capital exporting. The regressions in Table 1 include a linear year trend and a control for the gross domestic product (GDP) (measured in billions of $US). We obtain similar results when including country fixed effects, or when including a lagged dependent variable. We also obtain similar (and slightly stronger) results if we substitute the three-year moving average of the US interest rate as the key independent variable.

Notably, the negative relationship between interest rates and FDI inflows speaks only to the financial aspects of FDI, and does not necessarily imply that such a relationship exists with regards to commercial aspects of FDI (Kerner 2014).

More recently – and exotically – in early 2014 the Argentinean government announced a 50 percent tax on purchases over US $25 made by Argentinean customers on international websites, and it limited such transactions to at most two such purchases a year. These measures were, as observers noted, a deliberate “attempt to shore-up dipping reserves of foreign currency” (Garcia-Navarro, National Public Radio, 2014) and “to curb capital flight and prevent a possible balance-of-payments crisis” (Gilbert and Rathbone, Financial Times, 2014).

The ability to maintain interest rates at below world levels also allows governments to finance themselves domestically at cheaper rates, which is particularly valuable when foreign debt-loads are significant (Aizenman and Guidotti 1990).

It is also possible that creditors and credit rating agencies may favor countries that make durable, treaty-based commitments to the liberal economic order (Biglaiser and DeRouen 2007). BITs may provide an opportunity for countries to send such a signal (Büthe and Milner 2009; Kerner 2009) and may thereby ease access to other investment flows as well. While possible, Poulsen (2014) suggests good reason for skepticism in this regard.

It is important to point out explicitly that even the most fervent believer in BITs’ efficacy is unlikely to believe that the potential increases in capital inflows and export receipts would in themselves provide a solution to (potential or real) balance of payments issues. But to the extent that BITs do attract FDI, expand exports, or function as signals to creditors, they can very plausibly provide marginal help to countries during times of capital scarcity. To the extent that BITs do play such a role they almost certainly do so as a minor part in a larger set of liberal and illiberal policy initiatives meant to maintain access to hard currency. Our argument is simply that the intrinsic trade-off between the possibility of gaining access to foreign capital in exchange for policy autonomy will appear more palatable to governments at times when that capital is especially needed.

If the lender has taken on the exchange rate risk, devaluation of a country’s domestic currency will decrease the effective debt load (e.g., Walter 2008; Betz and Kerner 2014).

While some of the larger emerging markets have lessened the extent of their original sin in recent years, it remains a substantial problem even in the most developed of emerging market economies (e.g., Kynge 2015). This dynamic is reinforced by the fact that emerging markets typically borrow in short-term maturities (Broner et al. 2013).

Notably, this eliminates south-south BITs from our analysis.

The negative binomial model is a generalization of the Poisson model that allows for overdispersion in the data (Cameron and Trivedi 2005).

Nearly identical results when clustering standard errors by country or estimating these models with unclustered standard errors.

These results are also robust to controlling for the domestic interest rate.

See Betz (2014) for a different perspective on partisan hands-tying.

See Allison and Waterman (2002) for a discussion of fixed effects in negative binomial models.

These calculations omit China, which has substantially larger net assets than any other country in the sample. Including or omitting China from the sample does not matter for the substantive conclusions presented in the following, however.

It should be noted, however, that while our models provide evidence to suggest the existence of a range of low interest rates for which the effect of an increase in net foreign assets is to increase BIT signings and a range of high interest rates for which the effect of an increase in net foreign assets is to decrease BIT signings, our results give a substantially less consistent picture of how large those ranges are. The average effect, in any event, tends to be not statistically significant at conventional levels.

For better readability, the left panel of Fig. 1 omits the largest and smallest one percent of net foreign assets.

Several low-income countries had, at times, an interest rate of zero on newly issued government debt. Many of these countries obtained interest-rate free loans from the World Bank’s International Development Association. The following results are robust to omitting these countries from the sample.

A duration model has a number of advantages in the present context. Most importantly, we can include time-varying variables, such as US interest rates, in a straightforward manner (whereas with a standard linear regression model, we would have to use period-averages or another arbitrary value within the time period for such variables).

We include in our specification controls for GDP, GDP growth, the US inflation rate, and the US growth rate. The findings are robust to the inclusion of many other controls, including FDI and Trade as percentages of GDP, GDP per capita and others. These findings are also robust to excluding all controls.