The Great Recession and a Missing Generation of Exporters

Abstract

We study the impact of foreign market entry and exit by firms on the trajectory of U.S. exports during and after the Great Recession. Using confidential microdata from the U.S. Census Bureau, we find that incumbent exporters were primarily responsible for the changes in aggregate foreign sales during these years. While there was a substantial decline in the number of firms that sold abroad in the midst of the crisis, new exporters during the recovery compensated for this by having larger foreign sales. Thus, while changes in foreign market participation drove substantial shifts in the variety of U.S. goods that were exported, overall they were less important for the trajectory of total foreign sales over time.

Appendix

1.1 Aggregate Exports

Figure 7 shows the evolution of nominal aggregate exports during the Great Recession in comparison with the average trajectory during recessions over 1948–2001. Overall, this is similar to the results for real goods exports in Fig. 1 of the main text. As shown with the dashed line, historically these periods had only a minor impact, with exports staying roughly flat over the first eight quarters after the start of the downturn. The Great Recession, however, showed an initial increase in foreign sales over the first several quarters followed by a dramatic decline, falling to 20% below the initial level after six quarters. With regards to the recovery, exports initially rebounded faster after the Great Recession than they did in the past. Despite this, the recovery fails to catch up with the historical average by 2014.

Fig. 7

Nominal export growth after the Great Recession. Notes The figure shows the typical trajectory of log nominal goods exports 24 quarters after the most recent business cycle peak. This is done for the 11 recessions between 1948 and 2001 together as well as for the Great Recession

Figure 8 shows the evolution of goods exports prices during the Great Recession relative to the typical trajectory for recessions during 1948–2001. The periods used to construct this, as well as Fig. 1 of the main text, are 2007q4–2014q4, 2001q1–2007q1, 1990q3–1996q3, 1981q3–1987q3, 1980q1–1986q1, 1973q4–1979q4, 1969q4–1975q4, 1960q2–1966q2, 1957q3–1963q3, 1953q3–1959q3, and 1948q4–1954q4, in which the year is indicated and followed by the relevant quarter. Unlike in the past, the Great Recession initially brought along a temporary decline in export prices in the second year. Prices subsequently catch up with the typical trajectory before flattening out roughly 3 years after the peak.

Fig. 8

Goods exports price growth after the Great Recession. Notes The figure shows the typical trajectory of log goods exports prices 24 quarters after the most recent business cycle peak. This is done for the 11 recessions between 1948 and 2001 together as well as for the Great Recession. Prices for exported goods initially declined by 5% before rising almost 10%. Twenty-four quarters after the peak before the Great Recession, however, goods exports prices had increased only by about 7.5 log percent, which was roughly 5.5 lppt. less than after the average recession

1.2 Log Export Growth Decomposition

In this section, we present an exact decomposition of total export growth that follows di Giovanni et al. (2014). We begin by considering a three-term decomposition that is then extended into one with five terms. The log growth rate of total exports over h years $g_{t}^{h}$ can be decomposed into three additive terms as follows:

$$\begin{aligned} g_{t}^{h}= & {} \ln \left( X_{t}\right) -\ln \left( X_{t-h}\right) \nonumber \\= & {} \ln \left( \dfrac{\sum _{i\in I_{t\cap t-h}}X_{it}}{\sum _{i\in I_{t \cap t-h}}X_{it-h}}\right) +\ln \left( \dfrac{\sum _{i\in I_{t}}X_{it}}{\sum _{i\in I_{t\cap t-h}}X_{it}}\right) -\ln \left( \dfrac{\sum _{i\in I_{t-h}}X_{it-h}}{\sum _{i\in I_{t \cap t-h}}X_{it-h}}\right) \nonumber \\= & {} \underbrace{ \ln \left( \dfrac{\sum _{i\in I_{t \cap t-h}}X_{it}}{\sum _{i\in I_{t \cap t-h}}X_{it-h}}\right) }_{ i_{t}^{h}=\text {intensive margin } }+ \underbrace{ \underbrace{ \ln \left( 1+\dfrac{\sum _{i\in I_{t \setminus t-h}}X_{it}}{\sum _{i\in I_{t\cap t-h}}X_{it}}\right) }_{ e_{t}^{h}=\text {entry margin} }- \underbrace{ \ln \left( 1+\dfrac{\sum _{i\in I_{t-h \setminus t}}X_{it-h}}{\sum _{i\in I_{t\cap t-h}}X_{it-h}}\right) }_{ x_{t}^{h}=\text {exit margin} } }_{ n_{t}^{h}=\text {net extensive margin} } \end{aligned}$$

(9)

in which $I_{t} = \left\{ i : X_{it}>0 \right\}$ is the set of firms that export in year t. The set of firms that export in year t and in year t − h is given by $I_{t \cap t-h} = \left\{ i : i \in I_{t}, i \in I_{t-h}\right\}$ while the set that export in year t but not in year t − h is denoted by $I_{t \setminus t-h} = \left\{ i : i \in I_{t}, i \notin I_{t-h}\right\}$. Lastly, firms that export in year t − h but not in year t are in the set $I_{t-h \setminus t} = \left\{ i : i \notin I_{t}, i \in I_{t-h}\right\}$. Prior work, including Gopinath and Neiman (2014) and Kamal and Krizan (2012), has considered related decompositions.

The intensive margin $i^{h}_{t}$ in Eq. (9) is the log percentage point contribution to total export growth between years t and t − h and is defined as the log growth in sales by all firms that sold abroad in both years. We define the entry margin $e^{h}_{t}$ as the contribution to the growth of foreign sales by firms that exported in year t but not in year t − h. Similarly, we define the exit margin $x^{h}_{t}$ as the contribution to export growth of firms that sold abroad in year t − h but did not do so in year t. Foreign sales for any set of exporting firms can be written as the product of the number of exporters and average foreign sales per firm that are in the relevant set. For example, total exports in year t can be written as $\sum _{i\in I_{t}}X_{it}= N_{t}\left( I_{t}\right) {\bar{X}}_{t}(I_{t})$ in which the number of firms is $N_{t}(I_{t})$ and average exports per firm is ${\bar{X}}_{t}\left( I_{t}\right) =\frac{1}{N_{t}\left( I_{t}\right) }\sum _{i\in I_{t}}X_{it}$. Likewise, exports by incumbents can be written as $\sum _{i\in I_{t \cap t-h}}X_{it}= N_{t}\left( I_{t \cap t-h}\right) {\bar{X}}_{t}(I_{t \cap t-h})$.

Using the number of exporters and their average exports, the three terms in Eq. (9) can be decomposed further into the five terms in Eq. (4) of the main text as follows:

$$\begin{aligned} g_{t}^{h}= & {} \underbrace{ \ln \left( \dfrac{\sum _{i\in I_{t \cap t-h}}X_{it}}{\sum _{i\in I_{t \cap t-h}}X_{it-h}}\right) }_{ i_{t}^{h}=\text {intensive margin } }+ \underbrace{ \underbrace{ \ln \left( \dfrac{\sum _{i\in I_{t}}X_{it}}{\sum _{i\in I_{t\cap t-h}}X_{it}}\right) }_{ e_{t}^{h}=\text {entry margin} }- \underbrace{ \ln \left( \dfrac{\sum _{i\in I_{t-h}}X_{it-h}}{\sum _{i\in I_{t \cap t-h}}X_{it-h}}\right) }_{ x_{t}^{h}=\text {exit margin} } }_{ n_{t}^{h}=\text {net extensive margin} } \nonumber \\= & {} \underbrace{\ln \left( \dfrac{{\bar{X}}_{t}\left( I_{t\cap t-h}\right) }{{\bar{X}}_{t-h}\left( I_{t\cap t-h}\right) }\right) }_{i_{t}^{h}=\; \text {intensive margin }} \nonumber \\&+ \underbrace{ \underbrace{ \underbrace{\ln \left( \dfrac{N_{t}\left( I_{t}\right) }{N_{t}\left( I_{t\cap t-h}\right) }\right) }_{ee_{t}^{h}=\; \text {entry extensive }} + \underbrace{\ln \left( \dfrac{{\bar{X}}_{t}\left( I_{t}\right) }{{\bar{X}}_{t}\left( I_{t\cap t-h}\right) }\right) }_{ei_{t}^{h}=\; \text {entry intensive}} }_{ e_{t}^{h}=\; \text {entry margin } } - \underbrace{ \underbrace{\ln \left( \dfrac{N_{t-h}\left( I_{t-h}\right) }{N_{t-h}\left( I_{t\cap t-h}\right) }\right) }_{ xe_{t}^{h}=\; \text {exit extensive } } - \underbrace{\ln \left( \dfrac{{\bar{X}}_{t-h}\left( I_{t-h}\right) }{{\bar{X}}_{t-h}\left( I_{t \cap t-h}\right) }\right) . }_{ xi_{t}^{h}=\; \text {exit intensive } } }_{ x_{t}^{h}=\; \text {exit margin} } }_{ n_{t}^{h}=\; \text {net extensive margin}} \end{aligned}$$

(10)

This five term decomposition denotes the definition for average exports of firms in year t as ${\bar{X}}_{t}(I_{t})$, the average in year t − h as ${\bar{X}}_{t-h}(I_{t-h})$, and the average for firms in year t that sell abroad in both years as ${\bar{X}}_{t}(I_{t \cap t-h})$. Similarly, ${\bar{X}}_{t-h}(I_{t \cap t-h})$ denotes the average foreign sales in year t − h of firms that export in both year t and in year t − h. $N_{t}(I_{t})$ and $N_{t-h}(I_{t-h})$ are the number of firms that export in years t and t − h, respectively. By definition, the number of firms in year t that exported in both years is the same as the number that export in t − h so $N_{t}(I_{t \cap t-h})=N_{t-h}(I_{t \cap t-h})$. As such, the only way the intensive margin contributes to the total is through growth in average exports of firms that sell abroad in both years t and t − h.

Writing Eq. (10) using symbols for each margin gives

$$\begin{aligned} \underbrace{g_{t}^{h}}_{\text {total growth }} = \underbrace{i_{t}^{h}}_{\text {intensive}}+\underbrace{ee_{t}^{h}}_{\text {entry extensive}}+\underbrace{ei_{t}^{h}}_{\text {entry intensive}}-\underbrace{xe_{t}^{h}}_{\text {exit extensive}}-\underbrace{xi_{t}^{h}}_{\text {exit intensive}} \end{aligned}$$

(11)

in which total growth $g^{h}_{t}$ is in log percent growth since year t − h and the units of each of the terms on the right-hand side are contribution to total growth in log percentage points.

1.3 Additional Export Participation Regressions

Table 1 in the main text shows the effect of recessions on export participation. Table 8 extends this analysis and allows for differential impacts across the two recessions in our sample (2001 and 2007–2008). The main conclusion is that while both recessions had effects, the Great Recession had a larger one. This conclusion is consistent with the results in Table 2, which shows that export participation is positively associated with GDP growth. The Great Recession has a larger effect because it was more severe than the 2001 recession. Table 9 shows the unabridged version of Table 5 in the main text with coefficients for all covariates that are included in that regression. Table 10 provides summary statistics on our sample.

Table 8 Export participation and different recessions

Table 9 Export participation and GDP growth: durable goods, final goods, and high external finance dependence sectors

Table 10 Summary statistics

1.4 Sectoral Decompositions

Tables 11, 12, and 13 present the decomposition estimations from Table 7 of the main text for the manufacturing, services, and wholesale/retail sectors, respectively. All exports were deflated using the goods exports deflator from the national income and product accounts with 2000 as the base year. There are noticeable differences across the results for these three sectors. Manufacturing is by far the most influential sector in terms of aggregate exports. As a consequence, the results for manufacturing are quite similar to the results for overall aggregate exports shown in Table 7. At the same time, both services and wholesale/retail exports have been increasing in relative importance over the past couple of decades. In Tables 12 and 13, we see that both of those sectors grew at significantly faster rates than overall exports did during the 1993–2006 period. The 1-year log growth of services foreign sales averaged about 46%, whereas for wholesale/retail it averaged about 18%. This compares to roughly 6% growth for manufacturing and aggregate exports overall.

During the Great Recession, exports in manufacturing, in services, and in the overall economy declined by about 15% in the first year of the recession. In wholesale/retail, however, exports actually increased by about 72%. Six years after the beginning of the Great Recession, total exports and manufacturing exports on average grew about 2.4% per year, whereas in services they fell by about 11% per year and in wholesale/retail they rose by about 34% per year. With regards to the intensive and extensive margins, the picture is similar: Manufacturing exports behave like the aggregate figures, while in the services and wholesale/retail sectors, exports have higher growth rates along both margins. In particular, the wholesale/retail sector stands out as having the largest contribution from the net extensive margin.

Table 11 Real export log growth decomposition: manufacturing sector

Table 12 Real export log growth decomposition: services sector

Table 13 Real export log growth decomposition: wholesale and retail sectors

1.5 Partial Equilibrium Counterfactuals

1.5.1 Great Recession Versus Historical Averages

The detailed results for the counterfactuals shown in Fig. 4 can be calculated directly from Table 7. To do so, simply substitute the relevant counterfactual margin of interest from Panel A into the row of the same name in Panel B and then recalculate total growth in Panel B. While the Great Recession induced a missing generation of exporters, firms that started exporting tended to be bigger than they were historically. This fact muted the overall effect that the missing generation had on aggregate exports.

1.5.2 Allowing for Comovement with GDP Growth

Tables 14 and 15 present counterfactuals that allow various margins of export growth to follow their historical relationship with real GDP growth according to Eq. (8) in the main text. All exports are deflated using the goods exports deflator from the national income and product accounts with 2000 as the base year. Table 14 presents the results for exports across all sectors. If export growth followed its historical relationship with GDP growth, then it should have declined by 12 log percent by 2009 instead of declining by 16% as it did in the data (Table 7, Panel B, “Total Growth”). While the decline in exports was larger than expected, the recovery was significantly faster. In the 6 years following 2008, exports would have declined 3.3 log percent per year on average had they behaved relative to real GDP as they did in the past. In reality, exports grew 2.4 log percent per year and both the intensive and the net extensive margins outperformed their historical relationships with GDP growth. The intensive margin did better than the past by almost 5 log percentage points. In contrast, the net extensive margin contribution was only slightly larger than would be expected.

In Table 15, we separate the results for the whole economy shown in Table 14 into manufacturing, services, and wholesale/retail sectors. Comparing Table 15, Panel A to Table 11, Panel B, we see that manufacturing performed worse in the first year after the Great Recession in the data compared to what the past relationship would have suggested, but better after 6 years than what the historical relationship to GDP growth would have predicted. Better performance after 6 years in the data is led by the intensive margin, which performed about 5 log percentage points better than expected. The net extensive margin performed about 1 log percentage point better after 6 years.

Services exports fared much worse both 1 and 6 years after the Great Recession than the estimates with regards to their historical relationship with GDP growth in Eq. (8) implied. Those estimates would have predicted 27 log percent growth 1 year after the Great Recession and 20 log percent growth annually in each of the 6 years after (Table 15, Panel B). In contrast, services exports in the data fell 15 and 11 log percent 1 and 6 years after the Great Recession, respectively (Table 12, Panel B). At 5.5 log percent growth per year versus 44 log percent declines predicted by the model, the services sector had much faster intensive margin growth 6 years after the Great Recession than predicted. The net extensive margin, however, performed much worse in the data than predicted by the model. The entry intensive margin has the largest difference between the actual and predicted values. Six years after the Great Recession, this margin declined 1.6 log percent annually, but the model would have predicted an increase of 57 log percent. This implies that after the Great Recession, new exporters in the services sector were much smaller in the data than would have been predicted by the historical relationship with GDP growth.

For the wholesale/retail sector, our estimates in Table 15, Panel C, would predict 79 log percent growth when growth was actually 72 log percent after 1 year. Six years after, the predicted value is 42 log percent, whereas the actual one is 34 log percent. Overall, the actual value was lower than predicted at both horizons. Unlike the manufacturing sector, at the 6-year horizon this relatively worse performance is driven by the net extensive margin, which added 26 log percent in the model and only 20 log percent in the data. The entry margin appears to have been particularly important for the wholesale/retail sector. The exit margin contribution between the model and the data differed by about 2.2 log percentage points but at 8.2 log percentage points, the entry margin differed by substantially more.

Table 14 Real export log growth decomposition: allowing for comovement

Table 15 Real export log growth decomposition across sectors: allowing for comovement