Preparing to export

Danielken Molina, Marc Muendler 27 May 2013



Exporting is an essential feature of strategies for economic development for very good reasons. A large body of empirical evidence shows that exporters are larger, more productive, pay higher wages and hire more skilled workers (Bernard and Jensen 1995). But do firms move from local sales to export sales? What choices do firms make in preparation for exporting? How do these choices affect a firm’s future export performance?

New research on new exporters

In our recent work, we use data at the firm-worker level in Brazil and compare exporters to address these questions (Molina and Muendler 2013). We rank Brazilian exporters by their export-market participation over three consecutive years. We classify exporters as:

  • Export quitters – past exporters with a foreign shipment in at least one of the two preceding years but with no current foreign shipments.
  • Export starters – current exporters with no foreign shipment in at least one of the two preceding years.
  • Continuous exporters – current and past exporters with foreign shipments in three continuous years.

As Table 1 shows, Brazilian exporters differ considerably in performance between these categories. More persistent exporters are larger but exporters are not clearly distinguishable from each other in terms of observed workforce characteristics. Indeed, we document that more successful and larger exporters simply look like scaled-up versions of their smaller and less successful competitors, with similar workforce compositions.

Typically unobserved worker characteristics may be important determinants of expected export-market performance. In particular, we posit that firms actively prepare for expected exporting by hiring away a few key workers with prior experience at other exporting firms to gain their unobserved skills associated exporting. If true, this suggests that prior exporter experience of key workers may better predict export performance than conventional observable workforce characteristics.

Table 1. Export status ordering

aPermanent non-exporters do not export in any sample year; current non-exporters export in at least one sample year.

bNon-sustained exporters export in three consecutive years but serve no single destination in all three years; sustained non-OECD exporters serve at least one destination (but no 1990-OECD member country) in three consecutive years; sustained OECD exporters serve at least one 1990-OECD member country in all three years.

Source: SECEX 1990 through 2001 (t: 1992-2001), manufacturing firms (subsectors IBGE 2-13).

Notes: Universe of 1,767,491 manufacturing firm-year observations. Exports (fob) in thousands of August 1994 USD.

The Brazilian data

We use Brazil’s employer-employee linked data to track workers between employers from 1990 to 2001. To assess the importance of hires from other exporters, we relate the probability of exporting to an indicator of the export status of the former employers of new hires – including, of course, an exhaustive set of firm and workforce characteristics controls (see paper for details).

Our main finding is shown by Figure 1:

  • Hiring at least one worker with prior work experience at a previous exporting firm increases the probability of export-market participation by 2.8%.

This is a considerable probability shift, given an overall exporting frequency of only 5.5% in Brazilian manufacturing. It is similar in magnitude to what only substantive changes in observed workforce characteristics would predict.

For example, replacing the primary-educated share of the average workforce with tertiary-educated workers (which corresponds to an eightfold increase in the fraction of tertiary-educated workers from 0.1 to 0.8) raises the export probability by 2.5 percentage points.

Figure 1. Effects of hiring workers from an exporter and of a replacement of primary for tertiary-educated workers on export market participation.

Sources: SECEX and RAIS 1992-2001, manufacturing firms (subsectors IBGE 2-13).

Testing the main hypothesis

We then turn to tests of our main hypothesis that firms actively prepare for expected exporting by hiring a few key workers from other exporters. To construct a measure of expected export conditions one year in the future, we use current product-market conditions abroad. By moving the probability of exporting tomorrow, the exogenous product-market conditions abroad should exert a labour-demand shock that leads firms to actively prepare their workforces. Related earlier empirical research proceeded differently and asked how favourable labour supply conditions, for example the accession of a new manager with exporter experience, facilitated export performance at the current employer. Our paper poses the complementary question and asks conversely how favourable product-market conditions translate into a firm’s labour demand for skills pertinent to exporting.

Shifting attention from labour supply to labour demand allows us to make progress on statistical identification. Our new identification strategy is designed to assess export preparations in economically stable times. Concretely, we use current sector-level imports into destinations outside of Latin America from source countries other than Brazil (our instrumental variables) to predict a Brazilian firm’s future export status (our first stage). Then we regress the log of one plus the head count of hires from exporters (one plus the hired workers whose immediately preceding formal employment was at an exporter) on the predicted future export status (our second stage). The predicted future export status is a measure of a firm’s expected foreign market conditions, and our regression design allows us to test whether and how strongly expected exporting predicts the poaching of workers from exporters.

Figure 2. Log of hires from exporters as predicted with expected future export status

Sources: SECEX and RAIS 1992-2001, manufacturing firms (subsectors IBGE 2-13).

Note that on the left-hand side of the regression is the log of one plus the head count of hires from exporters. Results are robust to several alternative definitions of the left-hand side variable.

Figure 2 shows the main coefficient estimates. When we do not isolate the contribution of current market conditions for predicted export status (left bar in Figure 2), the measured effect of exporting a year later on hiring exporter workers in the present year is positive but relatively small. A ten-percentage-point increase in the chance of exporting one year later raises the number of hires from exporters by 1.2% today.

Once we predict future exporting with current market conditions, however, our approach isolates the causal effect of expected exporting on current poaching from exporters (right bar in Figure 2). Using the predicted future export status helps us measure the exogenous effect of a higher chance of exporting on the firms that are responsive to changing their workforce (the so-called responders in econometric terms). Future exporting causes significant advance hiring of former exporter workers.

The estimates from the instrumental variable regression are markedly higher in magnitude:

  • Increasing the probability of export-market participation next year by ten percentage points raises the number of hires from exporters by 19.1%;
  • At a sample mean number of poached hires from exporters of 1.1 workers per firm, a 10% higher export probability in a year leads to 0.4 more poaches from exporters this year.1

Firms in Brazilian regions with many exporters, large firms, and firms with lasting export-market participation are most responsive in hiring away other exporters’ workers. As theory suggests, firms with the largest anticipated gains from exporting have the strongest incentive to engage in preparatory hiring.

For example, we find that expected continuous exporters exhibit the strongest advance hiring. A corollary of our hypothesis is that a firm with favourable foreign market conditions but that fails to become an exporter will let go again of recent hires from exporters. In terms of our empirical strategy, we should find that a firm for which the current foreign demand conditions predict a high probability of export-market participation next year should lay off its recently poached hires from exporters when it fails to become an exporter. We find exactly that. Unexpectedly failing exporters lay off a significant fraction of their recently hired former exporter workers.

Figure 3. Future exporter performance as predicted by departing workers to other firms and hires of workers from exporters

Sources: SECEX and RAIS 1992-2001, manufacturing firms (subsectors IBGE 2-13).

In terms of performance (Figure 3), we document in our sample that firms with recently hired exporter workers exhibit both a wider reach of destinations (at the so-called extensive margin of exporting) and a deeper export-market penetration (at the intensive margin). We find these effects to be strongest when there is an overlap of export destinations between the former and the current employer. Poaching exporter workers in marketing-related occupations predicts a wider destination reach, whereas poaching skilled production workers predicts a deeper market penetration. These findings are consistent with the idea that exporters actively build up workforce expertise for expected export-market access. 

Why don’t firms fight poaching?

Why might firms threatened to lose a worker to a poaching firm not respond with a counter offer to retain the targeted worker? While we do not have conclusive evidence to answer this question, our results on firm performance provide a suggestion:

  • We find that exporters losing workers to other firms do not suffer a statistically significant decline in the number of export destinations, only a decline in market penetration.
  • Firms poaching workers from exporters experience improvement at both margins.

Firms losing a worker to other exporters may thus have a weaker incentive to retain the worker because the potentially prevented losses are smaller than the hiring firm’s expected gains.


Overall, the results are consistent with the idea that worker mobility is an important mechanism by which exporter knowledge spreads through the economy.

Disclaimer: The views expressed here are those of the authors and do not necessarily represent those of the institutions with which they are affiliated.


Bernard, Andrew B, and J Bradford Jensen (1995), “Exporters, Jobs, and Wages in US Manufacturing: 1976-1987”, Brookings Papers on Economic Activity: Microeconomics 1, 67–112.

Molina, Danielken, and Marc-Andreas Muendler (2013), “Preparing to Export”, NBER Working Paper, 18962.

1 Given the log of one plus the head count of hires from exporters on the left-hands side in the regression, we have .191*(1.1+1)=.401.



Topics:  Development

Tags:  Labour Markets, exports, firms

Trade Economist in the Integration and Trade Department, Inter-American Development Bank

Professor of Economics, Department of Economics at the University of California San Diego


CEPR Policy Research