Foreign-currency debt and the exchange rate pass-through

ABSTRACT

We show that higher foreign-currency indebtedness raises the degree of exchange rate pass-through to domestic producer prices. For identification, we use micro-level data from Turkey, an emerging market economy that has experienced large exchange rate movements over the last decade. Matching the Credit Register of Turkey with disaggregated manufacturing sector data on domestic prices and foreign-currency revenues from international trade, we show that sectors with higher ex-ante net foreign-currency liabilities raise their prices significantly more following domestic currency depreciation. The results are stronger if foreign-currency liabilities are short term.

KEYWORDS:

• Exchange rate pass-through
• producer prices
• foreign currency indebtedness
• emerging market economies

JEL CLASSIFICATION:

• E31
• F31

Disclosure statement

No potential conflict of interest was reported by the authors.

Disclaimer

The views expressed here are only those of the authors and should not be attributed to the Central Bank of the Republic of Turkey.

Notes

¹ As an emerging market economy, Turkey serves as an ideal laboratory to address this question. Its non-financial corporate debt-to-GDP ratio has raised over 30% after 2008 (Figure 1). The share of foreign-currency denominated debt in total debt has also been high compared to other emerging markets, reaching as high as nearly 60% in 2016 (Figure 2). Its foreign-currency debt-to-GDP ratio has increased by 20% points after 2008 (Figure 3). Finally, there exists cross-sectional heterogeneity among manufacturing sub-sectors in their foreign-currency indebtedness (Figure 4). We also would like to note that bank loans, rather than trade credit or bond issuance, are the primary source of foreign-currency funding for firms in Turkey, which also makes Turkey a convenient environment to study this question.

² Net FX liability is defined as total outstanding foreign-currency loans (including foreign-currency indexed domestic currency loans) of firms within sector $i$ minus net revenues from international trade (exports minus imports over the last 12 months); which in turn, is divided by total equity of firms in sector $i$. We use the CBRT Sectoral Accounts database that covers over 15,000 firms over our sample period. See Table 1 for average number of firms covered by this dataset for each sector.

³ Sectoral shares within the producer price index change mildly over the sample period. We take the time-average of each of these shares. See Table 1 for the sub-sector weights. The results are strongly robust to using unweighted (ordinary) least squares.

⁴ The Credit Register provides bank-firm-loan level details on outstanding credit balance, currency of denomination, maturity (short-term (<1 year) or >1 year), together with for which sector the loan is used. We aggregate outstanding foreign-currency credit balance at a three-digit sector level for short term and short-and-long term maturity loans.

⁵ The Central Bank of Turkey conducts a monthly survey on US dollar/Turkish lira (USD/TRY) exchange rate expectations. Survey results are publicly available at http://www.tcmb.gov.tr/wps/wcm/connect/en/tcmb+en/main+menu/statistics/tendency+surveys/survey+of+expectations Our sample period is from January 2007 to December 2016.

⁶ In particular, we had to leave aside sectors such as Leather and Leather Products (DC), Wood and Wood Products (DD), Coke, Refined petroleum products and nuclear fuel (DF), Electrical and Optical Equipment (DL), since the Credit Register does not provide further disaggregation regarding these sectors. Unless these sectors do not behave systematically differently in their within-sector pricing, e.g. ‘Wood’ compared to ‘Wood and Wood Products’ behaves systematically differently than, e.g., ‘Textiles’ compared to ‘Textiles and Wearing Apparels’, our results would continue to hold. Our sectoral coverage is 58% for 2007, and 68% for 2016.

⁷ If FX indebtedness were very strongly correlated with changes in the exchange rate, that would render our estimates hard to interpret. In this vein, we also find that our results are strongly robust to using average sectoral FX indebtedness, $\stackrel{‾}{{\left(\frac{\mathrm{N}\mathrm{e}\mathrm{t}\mathrm{F}\mathrm{X}\mathrm{L}\mathrm{i}\mathrm{a}\mathrm{b}\mathrm{i}\mathrm{l}\mathrm{i}\mathrm{t}\mathrm{y}}{\mathrm{E}\mathrm{q}\mathrm{u}\mathrm{i}\mathrm{t}\mathrm{y}}\right)}_{\mathrm{i}}}$, (not reported for brevity).

⁸ Throughout the text, we calculate the economic impacts by multiplying the estimated coefficient with the respective variables’ standard deviation. For instance, to reach the estimate of 2.2%, we multiply 0.017, the estimated coefficient, with the standard deviation of net foreign liabilities-to-equity ratio (which is 12.92 as given in Table 2) and by 10 (corresponding to the 10% increase in the exchange rate).