5.2 Long-run exchange rate pass-through

While short-run ERPT appears closely related to nominal price stickiness, firms may have more opportunity in the long run to adjust prices to reflect exchange rate movements. Gopinath et al. (2010) find that over the observed lifetime of US imported products, the gap between ERPT rates for USD and non-USD denominated contracts narrows, with pass-through to USD-denominated goods approximately doubling in the long run, to an average of 49 percent, while non-USD denominated goods maintain the very high rates of pass-through observed in the short run.[25]

Regressions include unreported HS4-destination fixed effects andmacroeconomic variables as outlined in the main text. Standard errorsin brackets (**; * denotes significance at the 1%; 5% level respectively). β coefficients all significantly different from each other at the onepercent level with the exception of β_producerand β_vehiclein column 3;and each of the β_vehicle coefficients with each other, and with β_producerwith in column 4.

Table 11 compares directly with the short-run calculation in Table 6 . In the long run, the average β across all currencies (0.256, column 1) is substantially lower than that observed in the short run (0.475), consistent with more flexibility in the long run to adjust prices to maintain constant NZD returns.

Allowing β to vary by invoice currency we see that the lower aggregate rate in the long run is driven by the same phenomena observed by Gopinath et al. (2010) - lower long-run βs (higher pass-through to import prices) in non-producer currency transactions (column 2). β for local currency pricers falls from 90.9 percent in the short run to 67.2 percent, while for vehicle currency pricers the drop is more dramatic, from 70.0 to 23.0 percent (column 3 of tables 6 and 11 respectively).

Table 12 compares directly with the short-run calculation in Table 8 . In the long run, when posted prices have more time to adjust, firm characteristics are somewhat more likely to have a distinguishable effect on ERPT, with six pairs of coefficients significantly different from each other at the five percent level or below. However, the extent to which characteristics matter for ERPT differs by invoice currency. Despite some sizeable differences in point estimates (β ranges from 0.385 to 0.755), there is no evidence that pass-through rates among local currency-invoiced trades differ significantly by firm characteristics. In contrast, for trades denominated in either the producer or vehicle currency, in many cases we cannot rule out the possibility of complete pass-through (β = 0) for high-performing firms. Although long-run βs are higher than the short-run βs among producer currency-invoiced trades overall, this effect is largely limited to firms with relatively low export performance. This may suggest selection at the extensive margin for smaller exporters, or a degree of market power among small, niche exporters, such that they are able to maintain their NZD returns in the face of exchange rate fluctuations. A similar pattern is observed among vehicle currency-invoiced trades. This may also be related to market power, or alternatively may suggest that some of the larger, vehicle currency exporters are in commodity sectors where currency movements are correlated with price movements (Chen & Rogoff 2003).

Finally, Table 13 returns to the Berman et al. (2012) style specification by holding β constant across invoice currency groups (as in Table 9 , the short-run equivalent), showing that many of the apparent differentials between firms with different characteristics wash out in the long run, as average βs across high and low groups converge. At the same time, estimated βs are lower across the board, due to the higher share of producer-currency pricing in the long-run sample (Table 14 ).

Regressions include unreported HS4-destination fixed effects and macroeconomic variables as outlined in the main text. Standard errors in brackets (** ; *denotes significance at the 1%; 5% level respectively). ‡; † signify that low-high coefficient pairs are significantly different from each other at the 1%; 5% levelrespectively.

Regressions include unreported HS4-destination fixed effects and macroeconomic variables as outlined in the main text. Standard errors in brackets (** ; *denotes significance at the 1%; 5% level respectively). ‡; † signify that low-high coefficient pairs are significantly different from each other at the 1%; 5%level respectively.