Abstract
We estimate an augmented gravity model using a firm-level database on Turkish firms to revisit the trade-exchange rate relationship over 2003–2015 at the intensive export margin. Besides several additional layers of analysis made possible by unique attributes of our firm-level database, we also examine exchange rate effects separately for firms engaged in manufacturing and services-intensive activities, which differentiates us from existing literature. A depreciation of the Lira is found to be associated with an increase in the quantity of Turkish exports, irrespective of firm ownership or size, though the effect is found to be muted for manufacturing firms in the sample that are more reliant on imported intermediates. Meanwhile, exchange rate volatility is found to be adversely associated with firm-level exports, especially for the sample of services-intensive and foreign-owned firms, irrespective of whether the Lira appreciated or depreciated. Finally, while the results display significant heterogeneity across sectors, at the regional level, they are driven by the location of production/export hubs in Turkey, namely İstanbul in Marmara, Ankara in Central Anatolia and İzmir in Aegean.
Disclosure statement
No potential conflict of interest was reported by the author(s).
Notes
1 At the same time, there are a few studies rejecting the negative relationship between exchange rate volatility and exports (Asseery and Peel Citation1991; Bailey, Tavlas, and Ulan Citation1986; Aslan, Çepni, and Gül Citation2021).
2 We also experimented with alternative lag structures in sensitivity analysis and found the results to be qualitatively similar.
3 Examining the role of forward participation in the exchange rate effect of Turkish exports needed more granular data in Turkey’s destination markets, which unfortunately was unavailable in our firm-level database. This said, the focus of our analysis is on the role of backward participation in exchange rate pass-through as it is the reliance on imported intermediates that provides a natural hedge against an appreciating currency. The forward integration channel is thus less consistent with our narrative.
4 We would like to thank an anonymous referee for this suggestion.
5 Unfortunately, we could not examine the hysteresis hypothesis empirically or even examine the trade effects of exchange rate changes and volatility at the extensive margin for computational-capacity-related constraints. For the extensive margin and hysteresis analyses, we needed to construct a data set that included observations for each conceivable f (firm), j (country) and t (quarter) triplet. When we built our extensive margin dataset, it became explosively huge and completely unwieldly: 41 countries × 52 quarters × 130,528 firms = 278,285,696 fjt observations. Even setting up the data in this manner took a long time. We also experimented with a few work-around strategies, by trying to collapse the quarterly data to annual data and trying to work with smaller sub-samples with quarterly data, but in vain.
6 The effects of exchange rate changes also depend on whether local currency pricing or producer currency pricing is used for exports and imports. Unfortunately, our firm-level database did not provide any information on the invoice currency to enable empirical analysis along this dimension.
7 The countries are Algeria, Austria, Azerbaijan, Belgium, Bulgaria, Czechia, China, Denmark, Egypt, France, Germany, Georgia, Greece, Hungary, Iran, Iraq, Ireland, Israel, Italy, Jordan, Kazakhstan, Lebanon, Libya, Morocco, Netherlands, Poland, Portugal, Romania, Russia, Saudi Arabia, Slovenia, Spain, Sweden, Switzerland, Syria, Tunisia, Turkmenistan, Ukraine, the UAE, the UK, the US.
8 We would like to thank an anonymous referee for this suggestion.
9 We only report results from the fully-specified preferred specification in all analyses hereinafter.