Sectoral Capital Flows: Covariates, Co-movements, and Controls

Abstract

This paper assembles a comprehensive sectoral capital flows dataset for 64 advanced and emerging economies from 2000-18, including direct, portfolio, and other investment to and from five sectors: namely, central banks (CB), general government (GG), banks (BKs), non-financial corporates (NFCs) and other financial corporates (OFCs). Using this data, the paper highlights the usefulness of a sectoral approach in assessing capital flow covariates, co-movements, and the effectiveness of capital controls. We show that 1) sectoral flows have varying sensitivities to measures of the global financial cycle and different cyclicality with respect to output growth; 2) co-movements in intra-sectoral resident and non-resident and co-movements with OFC sectoral flows explain a large part of the observed positive correlation between gross inflows and outflows; and, 3) sectorspecific tightening capital control measures appear effective in reducing the volume of flows to NFCs and OFCs.

Assessing the effectiveness of currency-differentiated tools

“…Overall, our results on side-effects highlight two main conclusions: first, FX reserve requirements appear to affect overall cross-border inflows beyond a simple reduction of domestic banks' external financing. This is consistent with results on the impact of currency based measures on capital flows (Ahnert et al, 2018 [5]; de Crescenzio, Golin and Molteni, 2017 [6]; Lepers and Mehigan, 2019 [4]; Frost, Ito and Van Stralen, 2020 [25]), and provides evidence of an impact on capital flows for FX reserve requirements. Secondly, while many studies have highlighted important circumvention of macroprudential tools, we do not see evidence of circumvention in the case of FX reserve requirements on the variables studied here.…”

“…Recent work, notably using credit registry or bank level data provides insights on some of the mechanisms: bank FX liabilities do feed credit dollarization (Özsöz, Rengifo and Kutan, 2015 [26]), with banks' non-core FX liabilities feeding credit dollarization three times more than core FX liabilities according to a study on Turkey (Yılmaz, 2020[27]). 4 The explanation may be related to the difference in maturity of non-core vs. core liabilities, the latter of which tend to be significantly shorter. 5 Overall, FX-differentiated reserve requirements are thus expected to help countries with issues of dollarization on both the asset and liability sides.…”