The Dual-Channel Transmission of Capital Flows: Evidence from the Taper Tantrum

1. Introduction

In May 2013, Ben Bernanke mentioned in congressional testimony that the Federal Reserve might consider slowing its asset purchase programme. Within weeks, capital had begun to exit emerging markets at scale. The episode entered the financial lexicon as the Taper Tantrum.

What makes the event analytically remarkable is not just its magnitude but its heterogeneity. The shock was identical for every recipient country: a single speech, on a single date, in Washington. Yet Turkey’s lira fell 15% against the dollar between May and December 2013. Poland’s zloty fell 3.5%. Indonesia’s rupiah fell 20%. Malaysia’s ringgit was barely affected. Brazil’s real fell 15% and required an emergency intervention programme worth $55 billion. Mexico initially depreciated sharply but recovered. This divergence between EMEs’ reactions is the puzzle at the centre of this paper.

The standard explanation points to macroeconomic fundamentals: countries with large current account deficits, high inflation, and thin foreign exchange reserves were more vulnerable. This is correct. Turkey had a current account deficit of 7.9% of GDP in 2013; Poland had a deficit of roughly 3 to 4%. But Poland had also been running current account deficits of 4 to 6% of GDP for the entire period 2006 to 2013. The 11-percentage-point gap in depreciation outcomes between Turkey and Poland is disproportionate to the current account gap alone. Something else was operating at a deeper level, in the structure of what had been accumulated during the QE cycle, not just in the size of the annual flow.

The central question of this paper: does the composition of the stock of external liabilities accumulated during an expansionary cycle determine both the nature of the credit expansion and the severity of the transmission of the shock that ends it?

The answer we argue is yes, and the mechanism is precisely the one described by Blanchard et al. (2015). We test this through a cross-sectional regression across twenty-eight emerging economies, expanding the empirical approach of Chari, Dilts Stedman, and Lundblad (2017).

2. Theoretical Framework

The standard open-economy framework, Mundell-Fleming and its descendants, predicts that capital inflows are broadly expansionary: greater external financing lowers the cost of capital, stimulates investment, and raises output. But this prediction rests on a simplifying assumption that becomes consequential in the emerging market context: the policy rate is held fixed. When monetary policy cannot or does not respond to the inflow, whether because the economy operates under an exchange rate anchor, faces trilemma constraints, or simply because the central bank chooses to hold, the inflow cannot be absorbed through rate adjustment. Instead, it discharges onto two variables: the exchange rate and the domestic cost of financial intermediation.

Blanchard, Ostry, Ghosh, and Chamon (2015) formalise this insight by distinguishing between two categories of domestic financial assets that foreign investors can acquire. Bonds are instruments whose return is tied to the policy rate. When foreign demand for domestic bonds rises, it appreciates the exchange rate but also raises the non-bond intermediation cost R_n​, since capital is being directed toward rate-sensitive instruments rather than into the broader credit system. Non-bonds (equity, cross-border bank loans, corporate instruments) are assets whose yield is not directly controlled by the central bank. When foreign demand for non-bonds rises, it similarly appreciates the exchange rate, but it compresses R_n : it reduces the cost of intermediation in the domestic economy independently of any policy rate movement. The two flows produce opposite movements in the credit channel, even when the aggregate inflow volume is identical.

Blanchard et al. (2015) Equilibrium Conditions (Simplified):