Productivity, Reallocation, and Economic Crisis: Evidence from Ecuadorian Firm-Level Data∗ Anson T. Y. Hoy Kim P. Huynhz David T. Jacho-Ch´avezx German Cubas{ Abstract The cleansing effect of recessions dictates that inefficient firms exit while only the most efficient firms enter. In a developing small open economy, distortions may prevent this efficient adjustment. Ecuador serves as an important case study because of its large eco- nomic crisis in the late 1990s. We study Ecuadorian firm dynamics to understand the sources of aggregate productivity growth. We find that reallocation of factor inputs and technical efficiency growth are the dominant factors. The reallocation due to net entry is only significant during the crisis period. These results illustrate the relative importance of extensive and intensive margins of reallocation. Keywords and phrases: Cleansing Effect, Aggregate Productivity, Input Distortions. JEL codes: D24, E25, L11, O11, O47. ∗We acknowledge the usage of the np package by Hayfield and Racine (2008) and the use of the Quarry High Performance Cluster at Indiana University where all the computations were performed. We thank Jason Allen, Calin Arcalean, Eric Bartelsman, Paul Carrillo, Kaiji Chen, John Earle, Oleksiy Kryvtsov, Jamil Mahuad, Os- waldo Molina, Mitsukuni Nishida, Amil Petrin, Robert Petrunia, B. Ravikumar, Bob Rijkers, Richard Rogerson, Kjetil Storesletten, Gustavo Ventura, Carolina Villegas-Sanchez, Kirk White, Yifan Zhang and participants of various workshops for comments. We thank Amil Petrin and Kirk White for sharing their productivity decompo- sition programs. The staff at the Ecuadorian National Statistics Office (INEC) provided great assistance with the data and answered many of our queries, in particular Galo Arias, Telmo Molina, Ver´onicaVel´azquez,Margarita Viera and Byron Villac´ıs. Ho thanks the U.S. Census Bureau for financial support. Jacho-Ch´avez thanks the Center for Latin American and Caribbean Studies (CLACS) at Indiana University for financial support. The views expressed in this paper are those of the authors. No responsibility for them should be attributed to either the Bank of Canada or the Bank of Uruguay. All errors are our own. yDepartment of Economics, Kansas State University, 327 Waters Hall, Manhattan, KS 66506-4000, USA. Email: [email protected]. zBank of Canada, 234 Wellington Ave., Ottawa ON, K1A 0G9, Canada. Email: [email protected]. xDepartment of Economics, Emory University, Rich Building 306, 1602 Fishburne Dr., Atlanta, GA 30322- 2240, USA. E-mail: [email protected]. {Bank of Uruguay and Department of Economics, FCS-UDELAR, Diagonal Fabini 777, Oficina 501, Monte- video, Uruguay CP 11100. Email: [email protected]. 1 1. Introduction Recent economic crises highlight the need for understanding the effects of a severe recession on developing small open economies. Ecuador's economic crisis of the late 1990's serves as an important case study of economic reallocation. Figure 1 illustrates the macroeconomic condi- tions faced by Ecuador in 1998-2000. One, the exchange rate suffered a massive depreciation in both nominal and real terms. Second, oil prices, one of the main exports, hit a record low (less than 10 US dollars per barrel). Third, the economy suffered massive capital flight due to the 1998 Asian financial crisis. During this crisis several financial and labour market reforms were undertaken. The rationale for these reforms were to remove impediments for the efficient allocation of economic resources. To understand the aggregate effects of a severe recession, we need to look at the micro evidence using firm level data because it allows to understand the inherent heterogeneity of firm behaviour. For example, Davis and Haltiwanger (1990) and Caballero and Hammour (1994) say that recessions, although painful, serve as a cleansing effect i.e. inefficient firms are culled while the efficient ones thrive. Using Ecuadorian firm level data from the annual survey of manufacturing and mining from 1998 to 2007, Figure 2 provides some salient facts regarding firm turnover and reallocation. One, job destruction was greater than job creation in both Light (labour-intensive economic sectors) and Heavy (capital-intensive) industries. Also, the firm exit rate was higher than firm entry rate in both industries. However, at least in the light industries entrants during this period were larger than the exitors. These graphs are suggestive of the cleansing effect of recessions. There is also an alternative view that recessions may have a negative reallocation effect. Barlevy (2002) observes that job quality is procylical and in the presence of job search frictions will generate low quality job matches. These low quality matches of firms and workers is an inefficient reallocation of labour and is known as the sullying effects of recessions. From a capital point of view, Ouyang (2009) shows that recessions may destroy potentially superior firms during their infancy, an effect known as scarring. These examples of market frictions prevent the efficient allocation of resources and is reflected in a distortion on the economy. These distortions may manifest themselves in either labour and/or capital inputs. We use the methodology developed by Restuccia and Rogerson (2008), Guner, Ventura, and Xu (2008), and Hsieh and Klenow (2009) to quantify input (capital/labour) distortions. Our results suggest that both the cleansing and sullying effect existed in Ecuador. In the labour market, the sullying effect is observed in both industries for most years. However, capital and labour input distortions fell temporarily after 2000 when the economy started to recover and substantial economic reforms took place. In those years, we find cleansing effects in the sharp decrease in labour and capital distortions. Distortions affect reallocation on the extensive and intensive margins. On the extensive 2 margin, distortions may prevent inefficient firms from exiting and efficient firms from entering, leading to a negative cleansing effect. On the intensive margin, distortions may prevent resources flowing from unproductive firms to productive ones as in a well-functioning economy. Petrin and Sivadasan (forthcoming) study Chile and find that labour market immobility could be a reason for large gaps in the value of marginal product of an input versus its marginal cost at any plant. Oberfield (2011) and Chen and Irarrazabal (2012) also find that misallocation has strong effects on TFP in the Chilean manufacturing sector. In terms of cross-country studies, Collard-Wexler, Asker, and Loecker (2011) find that much of the dispersion in TFP or capital misallocation is due to the dynamic adjustment costs of capital. Also, the presence of distortions may stimulate incumbent firms to improve their technical efficiencies to increase their chances of survival. With the existence of distortions, it is important to understand the relative importance of different effects and how they translate into aggregate productivity. To achieve this, we em- ployed the method developed by Petrin and Levinsohn (forthcoming) to decompose aggregate productivity growth (APG) into the effects of technical efficiency improvement, reallocation between incumbent firms (intensive margin), and reallocation due to entry and exit (extensive margin). Our results show that the dominant contribution to APG is a positive input reallo- cation, on average 4.9 and 6.4 percent for the Light and Heavy industries, respectively. This result reflects the substantial lessening input distortion during the economic crisis. Technical efficiency improved 4.6 and 3.7 percent for the Light and Heavy industries, respectively. Real- location on extensive margin or net entry was minimal, 0.1 and 1.8 percent for the Light and Heavy industries, respectively. Our study finds that reallocation has a cleansing effect on the intensive margin of reallocation. These findings are similar to Petrin, Reiter, and White (2011) who highlight the positive role of reallocation in the United States. Our findings contrast with previous work from Japan that indicates the natural selection mechanism may not lead to efficient reallocation, see Nishimura, Nakajima, and Kiyota (2005). Hallward-Driemeier and Rijkers (2010) document that in Indone- sia financial market imperfections attenuated relationship between productivity and survival. Understanding the underlying mechanism of APG, distortions and reallocation is important for policymakers as it quantifies on what margin (if any) policies should be directed towards. This paper is organized in the following fashion: Section 2 offers a background on the eco- nomic conditions during this period; section 3 describes the data used, offers some descriptive statistics, and investigates the reallocation and productivity patterns; section 4 analyzes the input distortions while section 5 discusses the APG decompositions; and section 6 concludes. 3 2. Background on the Ecuadorian Economic Crisis This section aims to provide a stylized summary about the currency crisis, and the subsequent official dollarization in Ecuador from 1998 to 2000. For a detailed analysis of the Ecuadorian currency crisis, see Beckerman (2002), and J´acome (2004).1 The Ecuadorian currency crisis originated from a series of external shocks. While agriculture products and crude oil were the major exports of Ecuador, El Ni~nofloods in late 1997 and 1998 destroyed vast agricultural areas in the coastal region and reduced Ecuador's agricultural productions. Oil prices in the world market also sank to a historical low - less than 10 USD per barrel - significantly