ABORDAJE DE LOS ENFOQUES DE EVALUACIÓN DE LAS POLÍTICAS PÚBLICAS

Relative to selling domestically, exporting activity is relatively rare. In various developed and developing countries, exporters represent just a tiny fraction of the total number of producers (Bernard et al. 2012). However, the differences between exporters and non-exporters are not random. A growing body of empirical evidence has shown that exporting firms are larger, more productive, and more skill- and capital-intensive than non-exporting firms. They also use a more varied input-mix and

pay higher wages.1

Two explanations for these differences have been proposed: self-selection and learning by exporting (LBE). The self-selection hypothesis suggests that the pre- conditions of exporters before exporting are already different compared to non- exporters. Exporters are already more productive than non-exporters, hence more able to make profits in the export market (Bernard, Redding & Schott 2007). Melitz (2003) develops a theoretical model showing that only firms with productivity above a certain ‘export productivity cut-off level’ find it profitable to export. This offers one explanation of the relation between productivity and export: more productive firms self-select into exporting.

The alternative hypothesis is that a firm becomes more productive after it engages in exporting. This second hypothesis, known as learning by exporting, postulates that when a firm breaks into an export market it obtains external knowledge from abroad, and this exposure allows it to improve its efficiency level. On the contrary, those that serve only domestic markets are denied such learning benefits

1 _{See Aw, Chung and Roberts (2000) for Taiwanese and Korean firms, Bernard and Jensen (1999, 2004),}

for USA firms, Bernard and Wagner (1997) for German firms, Clerides, Lach and Tybout (1998) for Colombia, Mexico and Morocco.

(Blalock & Gertler 2004). Evidence regarding this hypothesis is mixed. Aw and Hwang (1995), Bernard and Jensen (1999), Clerides, Lach and Tybout (1998), Delgado, Farinas and Ruano (2002) and Haidar (2012) find no learning effect from exporting. In contrast, Baldwin and Gu (2003), Blalock and Gertler (2004), De Loecker (2007) and Van Biesebroeck (2005) show some evidence of LBE. Martins and Yang (2009) conducted a survey analysis across countries and found that the impact of exporting on productivity tended to be higher in the case of exporters from developing countries than those from developed economies.

There are at least two channels through which LBE takes place; exporters can learn from buyers and competitors (Blalock & Gertler 2004). Buyers, especially of intermediate goods, may have the incentive to share knowledge, such as the latest design specifications and production techniques, as they want to obtain precise specifications and good-quality products. Meanwhile, more intense competition in foreign markets encourages firms to improve their efficiency and learn from competitors on how to survive in the markets.

Some studies suggest that export experience matters in determining future performance. Exporters often start by selling small quantities to a single neighbouring country. If this venture succeeds, they tend to keep exporting and start expanding to a new market and/or with new products (Albornoz et al. 2012; Álvarez, Faruq & Lopez 2013). Firms obtain more knowledge by exporting, and this generates persistence in exporting because profitability in the market rises with the length of export experience (Timoshenko 2015). Moreover, per period fixed costs, such as the costs involved in maintaining overseas distribution networks, are expected to fall as firms become more experienced and able to forecast foreign demand more accurately and find more

reliable overseas partners (Inui, Ito & Miyakawa 2016). All these studies show that export experience is a medium of learning.

If it is true that firms gain more knowledge through exporting, are older (or more experienced) exporters always more productive than the younger (or less experienced) ones? Blalock and Gertler (2004), using Indonesian data from 1990 to 1996, find evidence of LBE in Indonesia and show that a firm’s productivity increases by between 2 percent and 5 percent after it starts exporting. However, their study only compares productivity changes before and after exporting. De Loecker (2007) shows that Slovenian manufacturing firms are almost 9 percent more productive once they

start exporting and 13 percent more productive after four years of exporting.2 Alvarez

and Lopez (2005), using Chilean firm-level data, find that productivity gains from exporting take place only for new exporters and not for permanent exporters,

suggesting a short run effect of LBE.3

This chapter aims to provide an empirical analysis of how export experience can affect a firm’s productivity. Using firm-level data for Indonesian manufacturing over the period 2000–12, the study starts with some preliminary analyses to explore the behaviour of exporters and then employs several approaches to find evidence of LBE. In doing so, this study incorporates the export age as the proxy for experience. The preliminary analyses confirm that exporters in Indonesia are different from firms that serve only the domestic market; that there is evidence of a self-selection mechanism; and that export sales increase throughout the year of exporting. The main model in this study finds that productivity increases with export age, but the effect is not linear;

2 _{The impacts after exporting for five years and above are not significant due to decreased sample size.}

3 _{In Alvarez and Lopez (2005), permanent exporters are firms that always exported during the observed}

period.

it decreases once the firm becomes more experienced. Furthermore, LBE is more likely to be experienced by relatively larger firms and by those in certain industries in the unskilled labour-intensive sector. As far as I know, this is the first study that evaluates

the learning mechanism of Indonesian exporters using recent firm-level data.4

The remainder of this study is organised as follows. Section 2.2 briefly discusses the relevant literature. Section 2.3 describes the method and highlights some potential sources of bias. Section 2.4 explains the data and variables used. Section 2.5 discusses the results. Section 2.6 provides an array of robustness checks and, finally, Section 2.7 concludes the study.