persistencia de la rentabildad chile ppg114

5/24/2018 Persistencia de La Rentabildad Chile Ppg114

1/153

E - I S S N 2 2 4 8 - 6 9 6 8

I S S N 0 1 2 1 - 5 0 5 1

N N

V A

R

ESCUELA DE ADMINISTRACIN Y CONTADURA PBLICA

U N I V E R S I D A D N A C I O N A L D E C O L O M B I A

R E V I S T A D E C I E N C I A S A D M I N I S T R A T I V A S Y S O C I A L E S

S O C I A L A N D M A N A G E M E N T S C I E N C E S J O U R N A L

Enero-marzode2014,

Vol.24

nm.

51.

Valor$25.0

00.


2/153

Conten

J O U R N A

R E V I S T A

INNOVARREVISTA DE CIENCIAS ADMINISTRATIVAS Y SOCIALES Vol. 24, nm. 51 enero-marzo de 2014 ISSN0121-5051 E-ISSN2248-6968

UNIVERSIDAD NACIONAL DE COLOMBIA

Ignacio Mantilla PradaRector general/Chancellor

FACULTAD DE CIENCIAS ECONMICAS

Jos Guillermo Garca Is azaDecano /Dean

ESCUELA DE ADMINISTRACIN

Y CONTADURA PBLICA

Gloria Isabel Rodrguez LozanoDirector/Head of the School

MAESTRA EN ADMINISTRACIN

Luis Alejandro Rodrguez RamrezCoordinador acadmico/Academic coordinator

MAESTRA EN CONTABILIDAD Y FINANZAS

Germn Guerrero Chaparro

Coordinador acadmico/Academic coordinator

INNOVARREVISTA DE CIENCIAS ADMINISTRATIVAS Y SOCIALES

Dra. Gloria Isabel Rodrguez LozanoDirectora y editora general/Editor in Chief

PROCESO EDITORIAL/EDITING PROCESSEditor asociado/Associate editorLeonardo A. Paipilla PardoAlejandr o Pea Bello ([email protected])

Coordinadora editorial/Editorial coordinatorCarolina Gutirrez Rozo ([email protected])

Asistent e editoria l/ Editoria l Assista ntAna Ma. Castillo Montaa (revinnova_bog @unal.edu.co)

Correccin de estilo/Copyediting and ProofreadingEdwin Algarra Surez ([email protected])Edward Helbein ([email protected])

Traduccion es/Translatio ns:Proceditor Ltda. ([email protected])Roanita Dalpiaz ([email protected])

Diagramacin y diseo/Typesetting and design:Proceditor Ltda. ([email protected])

Ilustraciones/IlustrationsIvn Benavides Carmona ([email protected])

Impresin y acabados/Printing and final artProceditor Ltda. ([email protected])

SUSCRIPCIONES Y CANJE INTERNACIONAL:

Por favor comunquese con la Coordinacin de INNOVAR

al correo electrnico: [email protected]

Pgina web: www.innovar.unal.edu.co

Telfono (57) (1) 3165000 ext. 12308, 12367.

3 Editorial

Gestin y organizacin

5 Determinants of Working Capital Management in Latin American Companies

SAMUEL MONGRUT, DARCY FUENZALIDA OSHEE, CLAUDIO CUBILLAS ZAVALETA &

JOHAN CUBILLAS ZAVALETA

19 Relacin entre la creacin de valor y la inversin en I+D: unaaproximacin mediante redes neuronales artificiales

NGEL SAMANIEGO ALCNTAR & SAMUEL MONGRUT

31 Sistemas locais de produo: uma anlise dos fatoresinibidores das relaes de cooperao e interao

ANTONIO IACONO, MARCELO SEIDO NAGANO & EDMUNDO ESCRIVO FILHO

45 Una aproximacin holstica a la determinacin ygestin de precios en las compaas hoteleras

ROSARIO DEL CARMEN MARTN SAMPER & JOS MORENO ROJAS

Administracin pblica

61 Determinantes do endividamento da administrao local:estudo exploratrio nos municpios do norte de Portugal

NUNO ADRIANO BAPTISTA RIBEIRO & SUSANA MARGARIDA FAUSTINO JORGE

79 Gobierno corporativo y sector pblico: un estudiobibliomtrico en las principales revistas ISI

JESS MAURICIO FLREZ-PARRA, MARA VICTORIA LPEZ-PREZ &

ANTONIO MANUEL LPEZ-HERNNDEZ

99 El cambio estratgico en las empresas pblicas desde la perspectivadirectiva. Estudio de las televisiones regionales en Espaa

SILVIA SOSA-CABRERA & DOMINGO VERANO-TACORONTE

Economa y desarrollo

113 Es Chile un mercado competit ivo? Anlisis de la persistenciade la rentabilidad aplicando un panel dinmico

CLAUDIO ARAYA SASSI, PABLO RODRIGO & FRANCISCO MORALES

Finanzas

127 Los determinantes de la rentabilidad primaria de los bonos detitulizacin hipotecaria: la influencia de las estructuras multitramo

MIGUEL . PEA-CEREZO, ARTURO RODRGUEZ-CASTELLANOS &

FRANCISCO J. IBEZ-HERNNDEZ

Informacin editorial

143 Informacin editorial

145 Pautas para colaboradores

Innovar Vol. 24 nm. 51 incluye en su portada la obra deCarlos Jacanamijoy, Cadas de agua en Curiyacu. leo sobrelienzo. 1997.140 x 170.


3/153

NNOVAR

ESTRATEGIA Y ORGANIZACIONESAli Smida, Universidad Paris 13, Francia

Jos Ernesto Amors, Universidad del Desarrollo, Chile

Jos Gabriel Aguilar Barc el, Universidad Autnoma de Baja California, Mxico

Luis M. Silva Domingo,Universidad ORT, Uruguay

Vctor Ral Lpez, Universidad de Castilla La Mancha, Espaa

Ricardo Gouveia Rodrigues, University of Beira Interior, Portugal

Pablo Rodrigo Ramrez,Universidad Adolfo Ibez, Chile

Luis Arturo Rivas Tovar,Instituto Politcnico Nacional, Mxico

Gregorio Caldern, Universidad Nacional de Colombia(Sede Manizales), Colombia

Francisco Lpez Gallego, Universidad EAFIT, Colombia

CONTABILIDAD Y FINANZAS

Crawford Spence, United Arab Emirates University, Emiratos rabesJavier de Len Ledesma,Universidad de Las Palmas de Gran Canaria, Espaa

Jos Juan Dniz Mayor, Universidad de Las Palmas de Gran Canaria, Espaa

Mara Concepcin Verona,Universidad de Las Palmas de Gran Canaria, Espaa

Miguel ngel Martnez Sedano, Universidad del Pas Vasco, Espaa

Ramn Alfonso Ramos, Universidad de Santiago de Chile, Chile

Mary A. Vera, Universidad Nacional de Colombia, Colombia

Marysela Coromoto Morillo Moreno, Universidad de Los Andes, Venezuela

ECONOMA Y ORGANIZACIONESArturo Vsquez Prraga, University of Texas- Pan American,Estados Unidos de Amrica

Francisco Jos Delgado Rivero, Universidad de Oviedo, Espaa

GESTIN DE OPERACIONESGabriel Aramouni, Universidad de San Andrs, Argentina

Manuel Francisco Surez Barraza,Tecnolgico de Monterrey, Mx ico

GESTIN DEL TALENTO HUMANOLorenzo Revuelto Taboada, Universidad de Valencia, Espaa

MARKETINGAmparo Cervera Taulet, Universidad de Valencia, Espaa

Ana Isabel Jimnez Zarco, Universidad Oberta de Catalunya, Espaa

Hernn Talledo Flores, Grupo Episteme, Per

Joaqun Snchez Herrera, Universidad de Valencia, Espaa

Mara del Pilar Martnez Ruiz,Universidad de Castilla La Mancha, Espaa

Sergio Olavarrieta, Universidad de Chile, Chile

Sergio Moreno Gil, Universidad de Las Palmas de Gran Canaria, Espaa

Hctor Augusto Rodrguez Orejuela, Universidad del Valle, Colombia

Don Sexton, Columbia Business School,Estados Unidos de Amrica

COMIT EDITORIAL/EDITORIAL COMMITTEE

Dra. Gloria Rodrguez Lozano, Facultad de Ciencias Econmicas,Universidad Nacional de Colombia, sede Bogot

Dr. Francisco Rodrguez Vargas,Universidad Pars VIII, Francia

Dr. scar Castellanos, Facultad de Ingeniera,Universidad Nacional de Colombia, sede Bogot

Dra. Liliana Chicaza, Escuela de Administracin y Contadura Pblica,Universidad Nacional de Colombia, sede Bogot

Dr. Jorge Hernando Molano, Escuela de Administracin y ContaduraPblica, Universidad Nacional de Colombia, sede Bogot

Dr. Ernesto Sierra, Facultad de Ciencias Administrativas, Econmicas yContables, Universidad Central, Bogot

Dr. Carlos Alberto Rodrguez Romero, Escuela de Administracin yContadura Pblica, Universidad Nacional de Colombia, sede Bogot

Dr. Eduardo Senz Rovner, Escuela de Administracin y ContaduraPblica, Universidad Nacional de Colombia, sede Bogot

Dr. Carlo Tognato,The Universit y of Adelaide, Australia

Dr. Carlos Eduardo Maldonado, Universidad del Rosario, Bogot

Resumida, indexada o referenciada /Summarized, indexed or referenced: Sociological Abstrac ts (CSA) - USA Public Affairs Information Service (PAIS)- USA Thomson Gale. Informe Acadmico (Texto completo/Full text) - Mxico Cit as Latinoamericanas en Ciencias Sociales y Humanidades (CLASE) -Mxico Sistema Regional de Informacin en Lnea para Revistas Cientfica s de Amrica Latina, el Caribe, Espaa y Portugal (Latindex) - Mxic o Redaly- Mxico S ciELO Colombia - Colombia Directorio Ulrichs - USA ndice Nacional de Publicaciones S eriadas Cientficas y Tecnolgic as Colombianas(Publindex - Mxima categora A1) - Colombia Scopus (Elsevier B.V.) - USA.

El contenido de los artculos y reseas publicadas es responsabilidad de los autores y no refleja el punto de vista u opinin de la Escuela de Administracin deEmpresas y Contadura Pblica de la Facultad de Ciencias Econmicas o de la Universidad Nacional de Colombia. /The contents of all published articles andreviews are the authors responsibility and do not reflect whatsoever the point of view or opinion of the School of Business and Public Accounting of the Facultyof Economic Sciences or of the Universidad Nacional de Colombia.

El material de esta revista puede ser reproducido o citado con carcter acadmico, citando la fuente. /All published contents can be quote or reproduce foracademic purposes, acknowledging the source.

DIRECCIN INNOVARDirectora y editora general/Editor in Chief:Dra. Gloria Isabel Rodrguez Lozano

EDITORES CIENTFICOS /SCIENTIFIC EDITORS

COMIT CIENTFICO/SCIENTIFIC COMMITTEEDr. J. David Cabedo Semper, Universitat Jaume I de Castelln, Espaa

Dra. Amparo Cervera Taulet, Universidad de Valencia, Espaa

Dr. Sebastin Donoso, Universidad de Talca, Chile

Dra. Irene Gil Saura, Universidad de Valencia, EspaaDr. Javier Snchez, Universitat Jaume I de Castelln, Espaa

Dr. Eduardo Torres Moraga, Universidad de Chile, Chile

Dr. Gregorio Martn de Castro, Universidad Complutense de Madrid, Espa

Dr. Joaqun Alegre Vidal, Universidad de Valencia, Espaa

Dra. Teresa Garca Merino, Universidad de Valladolid, Espaa

Dra. Ma. Valle Santos lvarez, Universidad de Valladolid, Espaa

Dra. Mnica Gmez Surez, Universidad Autnoma de Madrid, Espaa

Dr. Jos Humberto Ablanedo Rosas, University of Texas at El Paso, USA

Dr. Lorenzo Revuelto Taboada, Universidad de Valencia, Espaa

Dr. Jos Solana Ibez,Universidad de Murcia, Espaa

Dr. Jos Enrique Devesa Carpio, Universidad de Valencia, Espaa

Dr. Javier de Len Ledesma,Universidad de Las Palmas de Gran Canaria, E

COLABORADORES EN ESTE NMERO/CONTRIBUTING AUTHORS FOR THIS ISSUE:

Samuel Mongrut Darcy Fuenzalida Claudio Cubillas Johan Cubillas ngel Samaniego Antonio Iacono Marcelo Seido Edmundo

Escrivo Rosario Del Carmen Martn Jos Moreno Jess M. Flrez-Parra Mara V. Lpez-Prez Antonio M. Lpez-Hernndez

Silvia Sosa Domingo Verano Claudio Araya Pablo Rodrigo Francisco Morales Nuno A. Baptista Susana M. Faustino Miguel A.

Pea-Cerezo Arturo Rodrguez-Castellanos Francisco J. Ibez-Hernndez


4/153

N

3REV. INNOVAR VOL. 24, N M. 51, ENERO-MARZO D E 2014

El primer nmero de 2014 de la revista Innovar abor-da temas de indudable inters en el mbito nacionale internacional, que se despliegan en cuatro seccio-nes: 1) Gestin y organizacin, 2) Administracin pblica,3) Economa y desarrollo, y 4) Finanzas.

En lo que respecta a la Gestin y organizacin, se presen-tan cuatro artculos: el primero de ellos, Determinants ofworking capital management in Latin America, nos mues-tra cmo algunos factores influyen sobre la manera en quelas compaas latinoamericanas gestionan su capital detrabajo; el segundo, intitulado Relacin entre la creacinde valor y la inversin en I+D: una aproximacin mediante

redes neuronales artificiales, analiza a travs de un mo-delo no lineal la relacin de creacin de valor y la inver-sin en Investigacin y Desarrollo (I+D); el tercero, llamadoSistemas locais de produo: uma anlise dos fatores ini-

bidores das relaes de cooperao e interao, es un art-culo que presenta factores organizacionales y ambientalesque inhiben las relaciones de cooperacin e interaccin del

Sistema Local de Produccin (SLP), mediante un estudioexploratorio en un sistema de tipo embrionario, que actaen el segmento de mquinas e implementos agrcolas, ubi-cado en la regin occidental de Paran, Brasil; el ltimo,intitulado Una aproximacin holstica a la determinaciny gestin de precios en las compaas hoteleras, es un casode estudio en el sector hotelero donde se hace la identifi-cacin de los factores que intervienen en la determinacinde las tarifas hoteleras y gestin de los mismos, sealandoque los principales factores que influyen en las decisionesde precio en la empresa hotelera son el nivel de deman-da, las percepciones de los clientes y las actuaciones de la

competencia.En cuanto a la Administracin pblica, el artculo Determi-nantes do endividamento da administrao local: estudo

exploratrio nos municpios do norte de Portugal,analizalos determinantes del endeudamiento de la administracinlocal en municipios del norte de Portugal, que es clave enla sostenibilidad financiera, evidenciando que tanto en unanlisis bivariado como en el multivariado los principa-les factores determinantes son los mismos: el lmite legalmximo de endeudamiento, los gastos totales, los ingresosde capital y el saldo presupuestal corriente. Otro artculo,

Editori

con el ttulo Gobierno corporativo y sector pblico: un es-tudio bibliomtrico en las principales revistas ISI, dirigido aanalizar el Gobierno Corporativo (GC) en el rea del sec-tor pblico, realiza un anlisis bibliomtrico comprendidoentre los aos 2000 y 2008, donde se analizan las tem-ticas ms relevantes y significativas, as como los autoresms citados y las teoras ms empleadas, poniendo de ma-nifiesto las principales lneas de trabajo desarrolladas en

un rea de estudio todava incipiente y adicionalmente vecmo los enfoques sobre la aplicacin de cdigos de bue-nas prcticas, tica corporativa y responsabilidad social di-fieren de los adoptados en el mbito privado. El ltimoartculo de esta seccin, intitulado El cambio estratgicoen las empresas pblicas desde la perspectiva directiva. Es-

tudio de las televisiones regionales en Espaa, propone unmodelo explicativo de la capacidad de cambio estratgicode la organizacin pblica desde la perspectiva directiva, apartir de la presentacin de un estudio de mltiples casos,realizado en cadenas de televisin regionales en Espaa.

Sobre la seccin Economa y desarrollo, la investigacin lla-mada Es Chile un mercado competitivo? Anlisis de la per-sistencia de la rentabilidad aplicando un panel dinmicoanaliza mediante el uso de una metodologa de panel di-nmico el nivel de persistencia de la rentabilidad de firmascompitiendo en Chile, situando al pas en un nivel de com-petitividad interna incluso menor al de pases de un nivelde desarrollo econmico inferior.

Para finalizar, en la seccin de Finanzas el artculo, deno-minado Los determinantes de la rentabilidad primaria delos bonos de titulizacin hipotecaria: la influencia de las es-

tructuras multitramo, analiza los determinantes de la ren-tabilidad primaria de los bonos de titulizacin hipotecaria(conocidos en la literatura como mortgage backed securi-ties, o MBS) emitidos en Espaa durante el periodo 1993-2011.

GLORIA I. RODRGUEZ L. Ph. D.

Directora y editora general - INNOVAR

Profesora asociada de tiempo completo

Escuela de Administracin y Contadura Pblica

Facultad de Ciencias Econmicas

Universidad Nacional de Colombia, sede Bogot


5/153

Editorial

4 REV. INNOVAR VOL. 24, NM. 51, ENERO-MARZO D E 2014

The first volume of Innovar in 2014 deals with topicsof unquestioned interest both for Colombia and inter-nationally. It is divided into four sections: manage-ment and organization, public administration, economyand development and, lastly, finance.

Four articles are included in the section on managementand organization. The first (Determinants of working capi-tal management in Latin America) discusses how certainfactors determine the ways in which Latin American com-panies manage their working capital; the second (The re-lationship between the creation of value and investment in

R & D: An approach employing artificial neural networks)employs a non-linear model (artificial neural networks) toanalyze the relationship between the creation of value andinvestment in R&D; the third article (Local Production Sys-tems: An analysis of factors inhibiting cooperation and in-

teraction) is a an exploratory study of an embryonic LocalProduction System (LPS) operating in the agricultural ma-chinery and implements sector in the west of Paran state,

Brazil. It presents the factors that inhibit the development ofcooperation relations and interaction of the LPS. The last ar-ticle in this section (A holistic approach to determining andmanaging rates in the hotel sector) is a case study of the ho-tel sector, which identifies factors affecting hotel rates andthe procedures that are followed to set them. The articleindicates that the principal factors influencing companyprice decisions are demand levels, client perceptions andactions taken by the competition.

One of the articles in the section on public administration(Determinants of indebtedness in local government admi-nistrations: An exploratory study of municipalities in Nor-

thern Portugal) analyses the determinants of indebtednessin local government administrations in northern Portugal,a key aspect of the analysis of the financial sustainability.

The article shows that in both bivariate and the multipleanalyses the principal causal factors are the same: The le-gal maximum level of indebtedness, total spending, capitalreceipts and the current account balance Another article(Corporate Governance and the public sector: A bibliome-tric study of the principal ISI journals), examines CorporateGovernance (CG) in the public sector, carries out a biblio-metric analysis covering the 2000-2008 period, examines

the most significant topics covered by the research, themost cited authors and the most frequent theoretical ap-proaches used, illustrating the principal lines of researchdeveloped up to now in a still-incipient area of research.Furthermore, the article illustrates the ways in which pu-blic sector approaches to codes of good practice, corporateethics and social responsibility differ from their equivalentsin the private sector. The final article in this section (Mana-

gement perspectives on strategic change in publicly-owned

enterprises. A case study of regional television channels in

Spain) proposes an explanatory model of the capacity forstrategic change in publically-owned enterprises from the

management perspective. It is illustrated using a multiplecase study of Spanish regional television channels.

One article Chile a competitive market? Analysis of persis-tence of profits using a dynamic panel approach) is pre-sented in the section on economy and development. Thisstudy uses a dynamic panel approach to examine the per-sistence of profits of companies operating in Chile, findingthat Chile is less competitive even than countries withlower levels of economic development.

We present a final article (Determinants of the primaryprofitability of mortgage-backed securities: The influence

of multiframe structures) which examines factors determi-ning the primary profitability of mortgage-backed securi-ties (MBS) issued in Spain between 1993 and 2011.

GLORIA I. RODRGUEZ L. Ph. D.Director and General Editor - INNOVAR

Full-time Associate Professor

School of Administration and Public Accounting

Economic Sciences Faculty

Universidad Nacional de Colombia, Bogot campus


6/153

5

J O U R N A L

R E V I S T A

INNOVAR

5

CLASIFICACIN JEL: G39, M16.

RECIBIDO: enero de 2011 APROBADO: abril de 2013.

CORRESPONDENCIA: Av. Epigmenio Gonzlez 500. Fracc. San Pablo, C.P.76130, Quertaro, Mxico.

CITACIN:Mongrut, S., D. Fuenzalida, C. Cubillas y J. Cubillas (2014).Determinants of working capital management in Latin America. Inno-var, 24(51), 5-18.

Determinants of WorkingCapital Management in Latin

American Companies

Samuel MongrutProfesor de la Escuela de Graduados del Tecnolgicode Monterrey, Campus Quertaro, Mxico.Profesor Asociado del Centro de Investigacion de la Universidaddel Pacfico (CIUP), Lima, PerCorreo electrnico: [email protected]

Darcy Fuenzalida OSheeDoctor en Ciencias Empresariales de la Universidad de Lleida, Espaa. Magster en

Economa de la Universidad Catlica de Chile e Ingeniero Civil Qumico de la UniversidadFederico Santa Mara, Chile. Profesor del Departamento de Industrias de la UniversidadFederico Santa Mara, Casilla 110VCorreo electrnico: [email protected]

Claudio Cubillas ZavaletaLicenciado en Economa de la Universidad del Pacfico, Per. Analista de FinanzasCorporativas. Centro de Investigacin de la Universidad del Pacfico (CIUP), Lima, PerCorreo electrnico:[email protected]

Johan Cubillas ZavaletaLicenciado en Economa de la Universidad del Pacfico, Per. Analista de Capital Privado.Centro de Investigacin de la Universidad del Pacfico (CIUP), Lima, PerCorreo electrnico: [email protected]

ABSTRACT:The aim of this study is to determine the factors that affect working capital manage-ment in Latin American companies. Using an unbalanced panel data analysis for companies quotedin five Latin American capital markets it is shown that companies in Argentina, Brazil, Chile andMexico are holding cash excesses, which could destroy firm value. Results show that the industrycash conversion cycle, the company market power, its future sales and country risk have an influenceon the way Latin American companies manage their working capital with significant differencesamong countries in the region.

KEYWORDS:Working capital management, international management.

Introduction

Firms face a number of important decisions in their current operations andone of these important decisions concerns the efficient management of li-quidity. This decision is critical, as it is the reason for which many firms getto bankruptcy. In order to understand how important it is to have good li-quidity management, one may refer to two North American supermarkets:Walmart and Kmart (Shin and Soenen, 1998). These two companies had

DETERMINANTES DE LA GESTIN DEL CAPITAL DE TRABAJO ENLATINOAMRICA

RESUMEN: El objetivo de este artculo es determinar los factores queafectan la gestin del capital de trabajo en las compaas latinoameri-canas. Mediante el uso de un anlisis de datos de panel no balanceadosde compaas cotizadas en cinco mercados de capital de Latinoamrica, sedemuestra que las compaas en Argentina, Brasil, Chile y Mxico t ienenexcesos de efectivo, lo que podra destruir el valor de dichas empresas. Losresultados muestran que el ciclo de conversin de efectivo de la indus-tria, el poder de mercado de la compaa, sus ventas futuras y el riesgodel pas, influyen sobre la manera en la que las compaas latinoameri-canas gestionan su capital de trabajo, con diferencias significativas entrelos pases de la regin.

PALABRAS CLAVE:gestin de capital de trabajo, gestin internacional.

FACTEURS DTERMINANTS DE LA GESTION DU CAPITAL DE TRAVAILEN AMRIQUE LATINE

RSUM: Lobjectif de cet article consiste dterminer les facteurs qui af-fectent la gestion du capital de travail dans les compagnies latino-amri-caines. En utilisant lanalyse de donnes de panel non mises en quilibrede compagnies reconnues dans cinq marchs de capital dAmrique latine,il est dmontr que les compagnies en Argentine, au Brsil, au Chili et auMexique ont un excs de liquide, ce qui pourrait dtruire la valeur de cesentreprises. Les rsultats montrent que le c ycle de conversion de liquiditsde lindustrie, le pouvoir de march de la compagnie, ses ventes futureset le risque du pays, influent sur la manire dont les compagnies latino-amricaines grent leur capital de travail avec des diffrences significa-tives entre les diffrents pays de la rgion

MOTS-CLS :gestion du capital de t ravail, gestion internationale.

DETERMINANTES DA GESTO DO CAPITAL DE TRABALHO NAAMRICA L ATINA

RESUMO: O objetivo deste artigo determinar os fatores que afetam agesto do capital de trabalho nas companhias latino-americanas. Me-diante o uso de uma anlise de dados de painel no balanceados decompanhias cotizadas em cinco mercados de capital da Amrica Latina,demonstra-se que as companhias na Argentina, Brasil, Chile e no Mxicono tm excessos numerrios, o que poderia destruir o valor dessas em-presas. Os resultados mostram que o ciclo de converso numerria da in-

dstria, o poder de mercado da companhia, suas vendas futuras e o riscodo pas, influenciam sobre a maneira na qual as companhias latino-amer-icanas administram seu capital de trabalho, com diferenas significativasentre os pases da regio.

PALAVRAS CHAVE:gesto de capital de trabalho, gesto internacional.



7/153



similar capital structures in 1994; however, Kmart had acash conversion cycle (CCC) of 61 days while Walmart hadone of 40 days. The cash conversion cycle (CCC) reflectsthe interval of time (days) required to convert a dollar in-vested in current assets in cash. It is calculated by addingthe average period to collect to the average inventory pe-

riod, and subtracting the average period of payment.As a consequence of this difference Kmart faced ad-ditional financial costs in the order of US$ 200 millionper year. Clearly, this situation was not sustainable be-cause poor working capital management eventually con-tributed to Kmarts bankruptcy. This example illustrateshow working capital management, a key variable for anefficient management of liquidity, could lead the firm tobankruptcy when it is poorly handed. For this reason, theanalysis of working capital management is critical, as thispractice encompasses a number of policies relative to themanagement of liquidity. Working capital managementprovides the firm with information on the liquidity neededto operate efficiently. When payables are due before col-lectables there is a liquidity problem and, in extreme cases,payments can be suspended, which eventually could leadto the companys financial distress.

Empirical evidence shows that working capital manage-ment in the United States (Moussawi R et al., 2006) hassignificant effects on companies with two different resultsin place. On the one hand, to hold an adequate level ofworking capital, that is to say, an adequate managementof the CCC, may generate a situation where the companyincur in lower financial expenses and maintain a stablegrowth.

On the other hand, there can be a negative effect on theaccumulation of working capital, because overinvestmentcould destroy the value of companies. An example fromMoussawi et al. (2006) showed that the relationship be-tween working capital management and the value ofNorth American companies over the recent past decadeswas negative, precisely due to this overinvestment.

The case of Europe is very similar. The results of the survey

conducted by KPMG (2005) reveal that 74% of the leadingcompanies admitted that capital management is very im-portant, and consequently, they have developed policies toimprove its management. However, the European settingshows two problems relative to working capital manage-ment. The first , is that they do not analyze the componentsof CCC as a whole (i.e., they do not integrate credit, cashand inventory policies with the process of sales projec-tions) quite the contrary, the working capital managementis carried out from different perspectives assigning un-equal objectives to each one of the policies and therefore

maintaining an inefficient management of working cap-ital. Only 33% of the European firms analyze CCC in anintegrated manner.

The second problem in European firms lies in the conflictbetween target financial ratios and CCC. The main objec-tive of 58% of surveyed firms is to keep the levels of sales

high in order to obtain greater profit to fulfill target finan-cial ratios. Furthermore, the survey showed that there isno good follow-up of the CCC, which means that workingcapital management could be more reactive to the circum-stances of the company rather than well planned and mon-itored. Working capital management in firms that operatein Latin America shows an evident backward step over thelast years. In the mid nineties it was common to obtainloans from abroad, which were channeled through localbanks. These loans were placed at very low rates, and pe-riods of 180 days on average.

Furthermore, these loans could be renewed, so that theywere substituted by a new loan. For this reason, theseloans were considered a permanent source of income.However, since 1997, there was a series of internationalfinancial crises, which generated an interruption of thesefinancing sources and companies had to face their obliga-tions by cutting back their operating investments. This ledto the financial distress of a great number of companies.Firms sold less and tried to finance themselves with theirsuppliers, who did not collect, and hence, were in no po-sition to give them extended periods of payment. This isone possible explanation why companies could have heldexcessive working capital in recent years so as to face thissituation of scarcity of suppliers credit.

The survey carried out by Payne and Bustos (2008) landedat the conclusion that companies in Latin American areinadequately managing their working capital. The uncer-tainty with regard to payments and collections as well asa poor implementation of sales projections has led firmsin Latin America to overinvest in working capital. How-ever, another possible explanation for this overinvestmentby firms in Latin America is related to the short term in-vestment horizon. The growth Latin American firms have

recorded is recent, and consequently investment policieshave not aimed at the long term, but at the short term(Mongrut and Wong, 2005).

This study seeks to analyze the working capital manage-ment by firms operating in Latin America. The purpose isto assess whether they are overinvesting in working cap-ital, and subsequently to identify which variables have animpact on their CCC. The study is divided into five partsincluding this introduction: The second part reviews theliterature and describes the working capital management


8/153

J O U R N A L

R E V I S T A

INNOVAR


policies in Latin America. The third part is devoted to findout whether firms in our sample from Argentina, Brazil,

Chile, Mexico and Peru keep deficit or excess in cash.

In the fourth section a second set of variables was usedto analyze the determinants of CCC. In the fifth part, theconclusions of the study are discussed, while at the sametime some policy recommendations for the managing ofworking capital in Latin America are proposed.

Theoretical Framework

The Cash Conversion Cycle (CCC)

Working capital are the funds used to operate in the shortterm. If receivables are postponed, there can be delays inpayments and these could be suspended causing a situa-tion of illiquidity for the firm.

In this context, CCC is an important tool of analysis thatenables us to establish more easily why and how the busi-ness needs more cash to operate and when and how it willbe in a position to refund the negotiated resources. Figure1 illustrates the CCC.

FIGURE 1:The Cash Conversion Cycle

Operating Cycle

Cash Conversion Cycle

BuyInventory

Payment ofInventory

Sales ofproducts

CashCollection

Source: KPMG (2005)

A business can generate losses during a number of dif-ferent periods, but it cannot go on indefinitely with poorCCC management. The activities that are directly relatedto CCC management are the following:

Determining the effective number of days to collectreceivables

Determining the inventory needs

Determining the future growth of sales


9/153



These activities must be integrated in such a way thatthe period of time in which the cash is not being used tofund the working capital is minimized. These three activi-ties are carried out through the implementation of threepolicies: credit policy, inventory policy and cash manage-ment policy. The first policy is responsible for planning, ex-

ecuting, and monitoring sales growth projections.In brief, the goal of working capital management shouldbe to minimize the CCC, without having a negative impacton the quality of its components. That is to say, it is just asbad to have a surplus of working capital as it is to have adeficit of working capital. It is worth stressing there is nospecific manual on how to manage working capital, sinceit depends to a very great extent on the specific circum-stances of each company.

Literature review

Existing literature on working capital seems to have lostpopularity after the glorious period of the sixties and theseventies, when most of the models of working capitalmanagement were developed.

Even though these models were not formulated in an in-tegrated manner, they were a very important discussiontopic given their direct effect on the value of firms. Shinand Soenen (1998) researched the relations between CCCand the profitability of the firms for a sample of companieslisted in the United States Stock Exchange, during the pe-riod spanning from 1975 to 1994. They found a significant

negative relationship between the value of the companiesand the CCC of the same companies.

In addition to this, Shin and Soenen (1998) intended tocome up with the determinants of working capital, andfound that its management is correlated in a positive wayto firm size. They also established that industry concen-tration does not affect working capital management andthat a greater compensation paid to the CEO of the firm,definitely improves the companys management of workingcapital. These results suggest that working capital man-agement has an important impact on the profitability of

the firms.

Deloof (2003), in turn, carried out research on the relation-ship between working capital management and the perfor-mance of Belgian companies. This author used a sample of1009 non-financial Belgian companies for a period of timefrom 1992 to 1996. He came across a significantly nega-tive relationship between gross profits and the average pe-riod of receivables, the average period of inventories, andaverage period of payables. The results suggest that themanagers could create value for stockholders if they were

to reduce the time periods of receivables and inventories,to reasonably minimum levels. These results show thatthere is a certain level of working capital that maximizesthe value of the firms.

Arcos and Benavides (2006) wanted to estimate the en-trepreneurial efficiency of a set of companies in the non-

financial sector in Colombia, for the period from 2001 to2004. The results obtained were consistent with similarstudies conducted abroad in that the CCC was inversely re-lated to the profitability of the companies, when measuredwith respect to the level of sales.

Lazaridis and Tryfonidis (2006) conducted a stat isticalanalysis of 131 firms in Athens for the period 2001-2004, and concluded that managers may create bene-fits for the companies if they manage an adequate levelof CCC and maintain each one of its components atan optimal level. They also detected a negative rela-

tionship between the companys working capital and itsprofitability.

Vlez-Pareja et al. (2009) carried out a study in Latin Amer-ican markets, in order to determine the market perceptionof the use of liquidity surpluses that remain invested incash and/or short term investments. The results confirmedthe agency problem with respect to the undistributed cashby the companies. Liquidity surpluses should be distrib-uted, because if they are not, they destroy the value offirms because the market attaches greater value to the ex-pected value of the share, than the flow of dividends paid

out in cash. However, this study has a shortcoming be-cause it includes the return of cash equivalents in the cur-rent assets, and therefore, excluded them from the workingcapital. Hence, if all cash flows were already distributed toshareholders, there are no more possible return from shortterm investments and the return from these cash equiva-lent securities represent a contradiction. In brief, wheneverfinancial analysts forecast free cash flows they should dis-tribute all cash flows available and reflect this in the finan-cial statements.

Analysis of working capitalmanagement in Latin America

Latin America has recorded a very dynamic developmentover the past years. It has been possible to observe highlevels of sales, which reveal growth rates with two digitfigures. Likewise, capital markets have become more in-tegrated to the world markets (Mongrut and Fuenzalida,2007). Even though financial integration is far from beingsubstantial, the cost of capital for investments has gonedown and, as a result of this, Latin America has witnessed


10/153

J O U R N A L

R E V I S T A

INNOVAR


the entry of foreign investors with direct investments(Fuenzalida and Mongrut, 2010).

Despite the remarkable growth Latin America has expe-rienced, the management of average working capital hasnot been adequate. The evidence shows the deteriorationof CCC (see Figure 2), due to several problems such as defi-

cient collection policies and poor implementation of salesprojections, among others. The results of these problemslead firms in Latin America to hold, on average, an excessof liquidity which losses the alternative return of short terminvestment (Payne and Bustos, 2008).

FIGURE 2: CCC in Latin America (in days)

-10 0 10 20 30 40 50 60

Argentina

Brazil

Chile

Mexico

CCC (Days)

2006 2005 2004

-20

Peru

Source: Adapted from Payne and Bustos (2008)

One sees that the CCC of Latin American companies haverecorded an average increase from 36.9 days in the year

2005, to 37.6 days in 2007. In 2006 one can see that, ex-cept for Peru and Argentina, there has been an increase inCCC in Mexico, Brazil and Chile. It is also worth pointingout the high volatility of the CCC shown in the three yearssummarized in Figure No. 2. Despite this situation, thereare companies that have experienced a good working cap-ital management, where the receivables period, as well asthe inventory turnover have decreased, while the averagepayables period has increased.

A comparison of the firms with the best working capitalmanagement and those firms that have an average man-

agement, reveals that the gap has expanded remarkably,reflected in an average overinvestment in working capitalof USD 46 billion in year 2006 (Payne and Bustos, 2008).

Do firms in Latin America have excess of cash?

This section provides an explanation of the methodologyused to find out whether the companies, on average, keepexcess of cash. This section is divided in two parts: in thefirst part , a model with three versions is introduced in orderto help us find out whether firms, on average, have an

excess of cash. In the second part, the sample is describedand the results discussed.

Working capital management and firm value

The general hypothesis this section is based on,is that, on

average, companies in Latin American countries hold anoverinvestment in working capital. Therefore, it is to expectthat the relation between investment in working capitaland the value of the firm is negative.

In order to find out the net effect of the investment inworking capital on the value of firms, the valuation modeldeveloped by Kaplan and Ruback (1995) and followed byMoussawi R. et al. (2006) is used:

PV CASHCCF t

rF t

t

0 0

11

( )= ( )+ ( )

+( )=

(0)

Where PVF(0) is the current value of the firm, CASH(0)isthe company current value of its cash assets and CCF(t)isthe capital cash flow that must be discounted at the unle-vered cost of equity (r). This equation could be expressedalternatively in the following way:

PV CASHOCF t INV t INV t

rF

L S

t

t

0 0

11

( )= ( )+ ( ) ( ) ( )

+( )=

(1)

Where the firm operating cash flow OCF(t) is equal to:

OCF t REV t EXP t OTH t ( ) ( ) ( ) ( )= +

And where REV t( )is equal to net revenues, EXP t( )is equalto cost of goods sold plus administrative and selling ex-penses plus taxes, and OTH t( )is depreciation plus amor-tization. INV t

L( ) is investment in long-term assets and

INV t S( )is investment in current assets (cash balances plus

account receivables plus inventory) less current operatingliabilities (accounts payables and accrued expenses). How-ever, what matters is the net operating working capitalexcluding cash balances. For this reason, we are going toestimate three versions of equation (1), with and withoutchanges in cash and cash equivalents and one with onlychanges in cash excluding cash equivalents. This will allowus to separate the effect of cash management from therest of the working capital management.

On the basis of the previous method known as the Cap-ital Cash Flow (CCF) Method or the Compressed AdjustedPresent Value, one can obtain the first order condition tomaximize the company value:

PV t

INV t

REV t

INV t

F

S S

( )

( )

( )

( )= =0 1If,


11/153



Hence, the company maximizes its value to the extent thata dollar invested in working capital is equal to a dollar insales. According to Moussawi et al. (2006), in order to ana-lyze the relationship between the value of the firm and itsworking capital management, one could use three versionsof model (1).

First version: changes in short term investments

include changes in currents assets

For purposes of the analysis, it is assumed that cash flowsgrow at a constant rate, so that model (1) may be rewrittenas equation (2):

PV t CASH t OCF t INV t INV t

r gF

L S( ) ( ) [ ( ) ( ) ( )]

( )= +

+ + +

1 1 1

(2)

Rewriting the equation yields:

PV t CASH t OCF t r g

INV tr

F

L

( ) ( ) ( )

( )

= + +

+

11

11

+

g

INV tr gS

( )11

The equation can be restated by means of the followingregression model:

E PV t X CASH t OCF t F( )( )= + ( )+ +( )+/

*b b b0 1 2

1

b b3 4

1 1INV t INV t L S

* *+( )+ +( )

(3)

Where: OCF t OCF t r g

*( ) ( )+ = +

1 1 1

INV t INV t r gL L

*( ) ( )+ = +

1 1 1

INV t INV t r gS S

*( ) ( )+ = +

1 1 1

The marginal effect of net working capital investment isequal to:

PV t

INV tOCF tINV t r g r g

F

S S

( )( )

( )( )+ = ++ 1 11 1 1

PV t

INV t

OCF t

INV t r gF

S S

( )

( )

( )

( )+ =

++

=

1

1

1 1

14

In this way, the optimal level of working capital is attainedwhen b4is equal to zero. Besides, given that r g( ) >

10, if

b4 0< would mean that there is overinvestment in net

working capital and if b4 0> would mean that there is un-

derinvestment in working capital. However, in this modelwe are not able to distinguish if there is underinvestmentor overinvestment in cash and cash equivalents.

Second version: changes in short term

investments are divided in changes in non-cashand cash and cash equivalent investments

The previous version does not establish any difference be-tween the effect of working capital and the effect of thechange in the level of cash and cash equivalents of thecompany. Therefore, following Moussawi et al. (2006),equation (2) could be written as follows:

PV t CASH t OCF t INV t

r g

IN

FL( ) ( )

[ ( ) ( )]

( )

[

= + + +

+

1 1

VV t INV t

r gns C( ) ( )]

( )

+ +

1 1

(4)

In equation (4) the investment in net operating workingcapital is separated in two parts: the investment in non-cash working capital INVns(t+ 1) and the investment incash and cash equivalents INVC(t+ 1). In this way, it willbe possible to identify the effect of investment in cash andcash equivalents.

Rewriting the above equation yields:

PV t CASH t OCF t r gF

( ) ( ) ( )= + +

1

1

+( )

+

INV t r g

INV t r gL ns

1 1

1 1

( )

+

INV tr gC

( )1 1

Using the regression version of the model:

E PV t X CASH t OCF t

INV t INV t

F

L S

( )( )= + ( )+ +( )+

+( )

+

/ *

* *

b b b

b b

0 1 2

3 4

1

1 ++( )

+ +( )

1 15

b* *INV t C

(5)

Where:

OCF t OCF t r g

*( ) ( )+ = +

1 1 1

INV t INV t r gL L

*( ) ( )+ = +

1 1 1

INV t INV t r gS nS

*( ) ( )+ = +

1 1 1


12/153

J O U R N A L

R E V I S T A

INNOVAR


INV t INV t r gC C

*( ) ( )+ = +

1 1 1

The marginal effect of investment in net working capitalis equal to:

PV t

INV t

OCF t

INV t r g r gF

S S

( )

( )

( )

( )+ = +

+

1

1

1

1 1

PV t

INV t

OCF t

INV t r gF

S S

( )

( )

( )

( )+ =

++

=

1

1

1 1

14

In the regression of the second version, the coefficientsb

4 0< and b

5 0< , are interpreted as overinvestment in

working capital and in cash, respectively.

Third version: changes in short term investments are

equal to changes in operating working capital [T3]

If we include changes in cash equivalents in the changesof current assets it means that they will be subtracted fromthe free cash flows. If all cash has been distributed amongshareholders then it is not possible to have any more returnfrom short term investments and this will create a con-tradiction because the investment in marketable securitiescould generate a return and if so, it should be subtractednet of taxes from the value of the company.

In order to avoid this shortcoming we decide to assumethat cash equivalents and their return are effectively dis-tributed among shareholders, which means that we must

exclude then from the changes in current assets and in-clude then in the free cash flow. Hence equation (5) will bewritten as follows:

E PV t X CASH t OCF t F( )( )= + ( )+ +( )+/

*b b b0 1 2

1

b b3 4

1 1INV t INV t L Wce

* *+( )+ +( )

(6)

In this expression, INV t Wce

( )+1 refers to the company shortterm investment without changes in cash equivalents (mar-ketable securities). Again, if b

4 0< would mean that there

is overinvestment in net operating working capital withoutcash equivalents and if b

4 0> would mean that there is

underinvestment.

It is important to state that equations (3), (5) and (6) willbe estimated using firm level data. Table 1 shows theproxies that one uses to calculate the variables to run re-

gression models (3), (5) and (6).

Sample and results

This study included information of non-financial companiesand non-utilities from Argentina, Brazil, Chile, Mexico andPeru for the period 1996-2008. Colombia and Venezuelawere excluded from the sample due to the scarcity of dataand the high volatility of the data available. The informa-tion about the companies was obtained from the financialdatabases Economatica and Bloomberg, and in certaincases, it was completed with public information availableat the stock exchanges of the corresponding countries. Forexample, in the case of Peru, the information contained inthe mentioned databases was complemented with infor-mation provided by the Lima Stock Exchange.

In some cases, companies were dropped from the samplegiven the high volatility of their data. In the case of Brazil,there was a smaller amount of data than for the othercountries because information on investments in fixed as-sets was available only from the year 2004 onwards.

In the five countries analyzed, firms that had very scarce

information were also left out of the analysis. This is thereason why the number of companies per country differsthroughout the years. Table 2 summarizes the number offirms analyzed in each quarter. Only in the case of Chilethe number of firms is the same during the study period. Itis important to point out that the selected companies arethe most representative ones of the corresponding StockExchange by market capitalization.

The methodology used to estimate equations 3, 5 and 6was through an unbalanced Panel data model given that in

TABLE 1: Proxies to estimate the variables of equations 3, 5 and 6

PVF Firm value Market capitalization + Total Debt + Cash

CASH Cash Cash + Cash equivalents

OCF EBIDTA EBIT + Depreciation and Amortization

INVL CAPEX Change in gross fixed assets

INVS Working capital Current Assets Current operating Liabilities

INVC Cash + Cash equivalents Change in Cash + Cash equivalents

INVwce Cash Change in Cash

Source: Authors


13/153



certain years there was no information available for firms.

We ran the Haussmann test (not reported) and obtained

that panel data with fixed effects was the best option, so

we use it. The next issue was to choose the proper esti-

mation method for the models.Initially we considered pool

estimation with ordinary least squares, but the high re-

maining volatility of the financial data rendered the mean

of the sample unrepresentative. Hence, we decide to use a

regression per quantile, which is a non parametric estima-

tion with respect to the median, not with respect to the

mean of the dependent variable. Furthermore, we correct

any problem of heteroskedaskicity presented in the results.

Table 3a (see Appendix) shows the results from estimating

the two versions of the first model (equations 3 and 5), and

Table 3b (see Appendix) shows the result from estimating

the third version of the model, equation (6). As we may

see the Earnings Before Interest and Taxes plus Depreci-

ation and Amortization (EBITDA) and the investments in

fixed assets are positively related with the dependent vari-

able in the five countries, which suggests that additional

EBITDA and investments in fixed assets would generate an

increase in the value for most of the companies.

According to the results, the firms operating in Peru are

the only ones that clearly are managing their working cap-

ital well. Chile is in a process of improvement, but in the

cases of Argentina, Brazil and Mexico the hypothesis of

overinvestment in working capital is verified. Furthermore,

it is important to note that the effect of overinvestment

in cash and cash equivalents is verified in three of the

countries studied (Brazilian companies could also be in-

cluded because they have overinvestment in working cap-

ital). This result is consistent with the results obtained by

Vlez-Pareja, et al. (2009) where the same problem arose

with respect to funds undistributed by the firms.

Determinants of the CashConversion Cycle (CCC)

It is important to identify which factors could have an ef-

fect in the cash conversion cycle (CCC). To accomplish this

goal, this section is divided in two parts: in the first part

explains the model to identify the factors, then the second

part describes the sample and explains the results.

Model for the CCC

The investment in working capital that a given manufac-

turing company has, is quite different to that of a services

company. Because of this, it is important to take into ac-

count the industry effect, so a positive relationship be-

tween the level of working capital of the industry and the

level of working capital of the company is expected to be

found (Hawawini, Viallet and Vora, 1986).

The size of the company is an important variable when it

comes to establishing a working capital policy. In principle,

larger sized companies ought to have a greater workingcapital given their higher sales. Despite this, larger sized

companies may also have a better relationship with their

suppliers, in this way, it is expected that they should main-

tain a lower investment in working capital. Therefore,

the effect that the size of the company may have on its

working capital is uncertain.

The proportion of tangible assets that is accounted for

by fixed assets exerts an impact on the working capital

and it differs across companies. For example, the inventory

TABLE 2:Average number of companies per year to estimate equations 3, 5 and 6

Year Argentina Brazil Chile Mexico Peru

1996 61 51 33 70 45

1997 64 59 38 77 52

1998 69 73 38 78 55

1999 76 76 38 78 56

2000 78 77 38 81 572001 78 77 38 83 60

2002 80 79 38 85 60

2003 80 79 39 85 60

2004 80 81 39 85 60

2005 80 81 39 85 60

2006 80 81 39 85 60

2007 80 81 39 85 60

2008 80 81 39 85 60

Source: Authors


14/153

J O U R N A L

R E V I S T A

INNOVAR


problems of a manufacturer of parts for motorcars willdiffer from those of a manufacturer of software. Added tothis, the problems of account receivables that these twocompanies may have will be different. It is expected thatcompanies that are more capital intensive may need todecrease their working capital with some exceptions when

the company has substantial market power with its pro-viders such in the case of supermarkets that may have neg-ative CCC.

A firms sales expectations have an influence on the in-vestment in working capital (Nunn, 1981), and they willalso affect the CCC because if a manager anticipates thegrowth of its sales, it is quite likely that it may increase theinvestment in inventories. As a result, it is likely that thefirm may also increase the use of loans.

Country risk is a variable which is expected to have a signif-icant impact on CCC. Country risk is being measured with

an index called Emerging Markets Bond Index (EMBI)that indicates the risk premium that international inves-tors must charge for investing in debt issued by public andprivate companies. In the face of a lower country risk, com-panies will increase their short and long term investmentsissuing debt because the risk premium they will have topay is smaller.

Finally, Moussawi et al. (2006) argue that the marketpower of the company allows it to have better relation-ships with its suppliers and clients providing it with advan-tages over its competitors. Hence, the more concentrated

the industry, the greater the companys influence will beover its CCC.

The following model (Moussawi et al., 2006) allows us toidentify how the aforementioned factors affect the com-pany working capital. In this case, the dependent variablewill be the cash conversion cycle (CCC).

CCC i t ACCC i t SIZE i t

PTA i t FGS i t

, , ,

, ,

( )= + ( )+ ( )

+ ( )+ +(

0 1 2

3 4 8))+ ( )

+ ( )+ ( )

5

6

HHI i t

EMBI t i t

,

,

(7)

Where sub-index i indicates the different companies ineach country and the sub-index t indicates the period.Table 4 shows the proxies used to calculate the variablesto run the regression model in equation 7.

Sample and results

In order to estimate equation 7, we include firms from Brazil,Chile, Mexico and Peru for the period 1998-2008, whichdo not belong to the financial and public sectors in thosecountries. In this case Argentina was excluded from theanalysis because the Buenos Aires Stock Exchange didntprovide a distribution of companies by industry. Besides,some key information for building the explanatory varia-bles was not found. For the remaining cases, the variableswere constructed on the basis of the industry classificationgiven by each countrys stock exchange. The data was ob -tained from Economatica and Bloomberg, and the selected

companies are the most representative ones within eachstock exchange according to their market capitalization.

The model to estimate was an unbalanced panel datamodel with fixed effects since information was not avai-lable for all companies in all years. The necessary cor-rections for autocorrelation were also made. Wooldridge(2002) developed a very flexible test in which the lack ofautocorrelation is the null hypothesis of this test. If thistest is rejected, one may suspect the presence of autocorre-lation. This test was applied to each regression and only inthe case of Mexico we encountered a problem of this type.

The other problem to be dealt with was heteroskedasticity.For the purposes of this study, we applied Walds modifiedtest for heteroskedasticity, as this test operates even whenthe assumption of normality with respect to the errors isnot fulfilled. The null hypothesis of this test is that thisproblem i

2 2= does not exist for all the i= 1N, whereN is the number of companies.

When this null hypothesis is rejected, a problem of hetero-skedasticity exists. Once again, the test was applied to thecountries selected and this problem was found in the four

TABLE 4: Determinants of the Cash Conversion Cycle (CCC)

CCC Cash conversion cycle Average collection period + Average period of inventories + Average payable period

ACCC Average Industry CCC Average of cash conversion cyc le of the companies belonging to the same industry

SIZE Firm size Logarithm of sales

PTA Proportion of tangible assetsRatio of tangible fixed assets (property, plant, machinery and equipment, among others) / Total

assets

FGS Future sales Percentage growth of sales from two years in the future

HHI Herfindahl- Hirschman Index Ratio of firms sales / Industrys total sales

EMBI Country risk EMBI of the country

Source: Authors


15/153



countries. Once the problems mentioned earlier were iden-tified, adequate solutions were implemented in each case.In the case of Brazil, Chile and Peru feasible generalizedleast squares estimators were used because they only pre-sented heteroskedasticity. In the case of Mexico, which alsopresented the autocorrelation problem, one used the Prais-

Winsten regression for an unbalanced panel data model.Table 5 (see Appendix) shows the results of estimatingequation 7. In two out of the four countries analyzedthe market power of the companies seems not to playan important role in the CCC. This is due to a very highcorrelation between the size of the company and the Her-findahl-Hirschman index (not reported). Therefore, by in-cluding the variable size both effects are captured. In thissense, it can be stated that market power is a variable thataffects CCC as long as the firm is large. This result sug-gests that to the extent that a big company has greatermarket power, it will have a lower CCC.

The results show that the size of the company has a sig-nificantly negative relationship with the CCC because theymay have better relationships with their suppliers. In ad-dition to this, larger firms in Latin America are subject toless financing restrictions since they have a greater accessto the financial market at lower financial costs than theirsmaller counterparts.

The practices adopted by the industry also determine theworking capital policy. In all countries the average industryCCC is significant and has a positive relationship with the

companys CCC, which means that companies try to stayclose to industry policy. The expectation of sales growth ispositively associated with the companys CCC. As a rule,companies increase their investment in inventories when-ever they anticipate a growth in their sales.

Although it was not significant in three out of the fourcountries, country risk is an important variable in Brazil be-cause in the other countries, where a lower country risk isfaced, company uncertainty with respect to market beha-vior is lower, and consequently, the investment in workingcapital increases, resulting in CCC increases as well. Fi-nally, the results show that the companies which are moreintensive in tangible assets, seek to reduce the investmentin working capital, which means that their CCC will de-crease too.

Conclusions

The initial hypothesis was that Latin American compa-nies had recorded an overinvestment in working capitalduring the last decade given their short term investmenthorizon. In order to verify this hypothesis, three versions

of the first model were analyzed. As a result, one couldverify this overinvestment hypothesis in three out of thefive countries studied (Brazil, Chile and Mexico). However,in the two remaining countries, only in the case of Peru itbe argued clearly that there is evidence of an improvementin working capital management.

Given the importance of working capital management, westudied also the determinants of the companies Cash Con-version Cycle (CCC). The results show that the CCC is neg-atively correlated with firm size and positively correlatedwith the industry concentration index, which suggests thatLatin American companies are using market power to re-duce their CCC.

Similarly, practices adopted by the industry are very impor-tant in determining the companies level of working cap -ital. It was also verified that companies tend to invest inworking capital when they anticipate a sales growth and

companies in Brazil invest more in working capital when-ever there is a lower country risk following in this way awindow opportunity as it is also the case with the Ini-tial Public Offerings issued by Latin American companies(Mongrut, Valenzuela and Garay, 2009).

On the basis of these results and the empirical evidenceobtained in prior studies, it can be argued that it is nec-essary to conduct a more exhaustive follow-up study ofthe working capital policy in Latin American companies inorder to formulate a strategy to help companies avoid thechronic problem of overinvestment that could destroy their

firms value.To this end, Anand and Gupta (2002) made a proposal todevelop a monthly report of the main ratios that are partof the CCC and make a ranking of the companies by sec-tors. This type of ranking could foster financial managersto keep their level of working capital within reasonablebounds.

ReferencesAnand M. & C .P. Gupta (2002) Working Capital Performance of Corpo-

rate India: An Empirical Survey for the Year 2000-2001. Manage-ment and Accounting Research, 2(1), 43-67.

Arcos M. & J . Benavides (2006). Efecto del ciclo de efecti vo sobre larentabilidad de las firmas Colombianas. Borradores de Economa

y Finanzas9, Cali, Colombia, Ediciones ICESI, 19.

Deloof M. (2003). Does Working capital management Affect Prof-itability of Belgian Firms?Journal of Business Finance and Ac-counting,30(3-4), 573-588.

Fuenzalida D. & S. Mongrut (2010). Estimation of Discount Rates inLatin America: Empirical Evidence and Challenges.Journal of Eco-nomic, Finance and Administrative Sciences, 15(28), 7-43.

Hawawini G., C. Viallet, &A. Vora (1986) Industry Influence on Cor-porate Working Capital Decisions. Sloan Management Review,27(4)15-24.


16/153

J O U R N A L

R E V I S T A

INNOVAR


Kaplan S. &R. Ruback (1995). The Valuation of Cash Flow Forecasts: AnEmpirical Analysis.Journal of Finance, 50(4), 1059-1093.

KPMG (2005) Working capital management: How do European Compa-nies Manage their Working Capital? KPMG Survey, Advisory, LLP.

Moussawi R., M. La Plante, R. Kieschnick &N. Baranchuk (2006). Cor-porate Working capital management: Determinants and Conse-quences. Working Paper. University of Texas at Dallas.

Lazaridis I. & D. Tryfonidis (2006). Relationship between Working cap-ital management and Profitability of Listed Companies in theAthens Stock Exchange . Journal of Financial Management andAnalysis, 19(1), 32-59.

Mongrut S., A. Valenzuela & A. Garay (2009) What Determines GoingPublic in Latin America? Chapter 13, 255-271. En, Emerging Mar-kets: Performance, Analysis and Innovation, G. N. Gregoriou (Ed.),Chapman-Hall/Taylor and Francis.

Mongrut S. & D Fuenzalida (2007) Valoracin de inversiones reales enLatinoamrica: hechos y desafos.Academicos, 2(4), 67-80.

Mongrut S. and D. Wong (2005) Un examen emprico de las prc-ticas de presupuesto de capital en el Per. Estudios Gerenciales,20(95), 95-111.

Nunn K. (1981). The Strategic Determinants of Working Capital: AProduct-Line Perspective. Journal of Financial Research, 4(3),207-219.

Payne S. & K. Bustos (2008). Latin America companies holding up toUS$ 46 Billions in working capital. REL/CFO Magazine ,4(1), Feb-

ruary, 1-4.

Shin H. &L. Soenen (1998). Efficiency of Working Capital and Corpo-rate Profitability.Journal Financial Pract ice and Education, 8(2),37-45.

Vlez-Pare ja I., M. Merlo, D. Londoo, & J. Sarmiento (2009) PotentialDividends and Actual Cash Flows: A Regional Latin American

Analysis . Estudios Gerenciales. Journal of Management and Eco-nomics of Iberoamerica, 25, (113), 151-184

Wooldridge J. (2002). Econometric Analysis of Cross Section and PanelData, Cambridge, MA, MIT Press, 776.


17/153



Appendix

TABLE3A:ExcessofcashinLatinA

mericanCompanies

Thesearetheresultsofestimatingequations3and5usingaPanelDatamodelwithfixedeffectsandquantileregression.

PV

F

isthedepe

ndentvariableandrepresentsthevalueofthefirmattimet,calculatedas

thecompanymarketcapitalization,lessTotalDebt,plusCash.CASHrepresentsthelevelofaccumulationofcashandcashequivalentsof

thecompanyintimet.

OCFrepresentsthefirms

EBITDAofyeart+1.INVL

representstheinvestmentinfixedassetsmadeduringyeart+1.INVSrepresentstheinvestmentinnetworkingcapital,wherenetworkingc

apitalisdefinedascurrentassetsminuscurrentoperatingliabilities(period

t+1minust)andlesscash(equation5).IN

VCrepresentstheinvestmentmadeincashandincashequivalentsduringtheyeart+1.

Equation3

Equation5

Variable

Argentina

Brazil

Chile

Mexico

Peru

Argentina

Brazil

Chile

Mexic

o

Peru

Constant

2283.67**

4500

00000***

-7819.93***

13866***

12543.99***

-5952.2***

55

3000000***

-9352.62***

79.19***

10161.6***

p-value

0,05

0,00

0,00

0,00

0,00

0.03

0.00

0.00

0.97

0.00

CASH

1.8793***

-87

52.2***

1.1841***

1.0928***

0.5313***

2.24289***

-16825.3***

1.27674***

0.98926***

0.71792***

p-value

0.00

0.00

0.00

0.00

0,00

0.00

0.00

0.00

0.00

0.00

OCF

19.0132***

200

05.6***

31.618***

20.7797***

16.0681***

18.4016***

35736.3***

31.6668***

21.0234***

16.3092***

p-value

0.00

0.00

0.00

0.00

0,00

0.00

0.00

0.00

0.00

0.00

INVS

0.3134***

-9295.52***

0.2512***

-0.3079***

4.4161***

-0.0635***

-3294.03**

-0.4716***

-0.06469

***

4.22492***

p-value

0.00

0.00

0.00

0.00

0,00

0.01

0.06

0.00

0.00

0.00

INVL

5.2757***

218

8.73***

10.5606***

12.7767***

3.6942***

5.75019***

2053.09*

10.7696***

12.3063***

3.13174***

p-value

0.00

0.00

0.00

0.00

0,00

0.00

0.08

0.00

0.00

0.00

INVC

-1.6784***

-1.2515***

1.17636***

-1.6006***

0.62336***

p-value

0.00

0.00

0.00

0.00

0.00

Obs.

2267

270

1274

3230

1045

2259

270

1272

3226

1045

Pseudo-R2

0.71

0.37

0.61

0.65

0.55

0.72

0.32

0.61

0.65

0.56

***Significantat99%levelofconfidence,**Significantat95%levelofconfidence,*Significantat90%levelofcon

fidence


18/153

J O U R N A L

R E V I S T A

INNOVAR


TABLE 3B: Excess of cash in Latin American Companies

These are the results of estimating equation 6 using a Panel Data model with fixed effects and quantile regression.PVFis the dependent variable andrepresents the value of the firm at time t, calculated as the company market capitalization, less Total Debt, plus Cash. CASH represents the level of ac-cumulation of cash and cash equivalents of the company in time t. OCF represents the firms EBITDA of year t+1. INVLrepresents the investment in fixedassets made during year t + 1. INVwcerepresents the investment made in cash, without cash equivalents, during the year t+1.

Equation 6

Variable Argentina Brazil Chile Mexico Peru

Constant -18828*** -206000000*** -484.7964 143818.6*** 5508.511***

p-value 0,00 0,02 0,85 0,00 0,01

CASH 1.9518*** 18019.53*** 2.2945*** 1.3463*** 0.67300***

p-value 0,00 0,00 0,00 0,00 0,00

OCF 18.971*** 12352.27*** 32.9123*** 19.7982*** 16.5529***

p-value 0,00 0,00 0,00 0,00 0,00

INVL 5.566*** 51843.11*** 1.5594*** 0.62250*** 2.0072***

p-value 0,00 0,00 0,00 0,00 0,00

INVwce 0.283*** -68835.29*** -0.9566*** -1.2168*** 3.7493***

p-value 0,00 0,00 0,00 0,00 0,00

Obs. 2238 277 1520 2198 1041

Pseudo-R2 0,72 0,67 0,51 0,65 0,53

*** Significant at 99% level of confidence, ** Significant at 95% level of c onfidence, * Significant at 90% level of confidence

TABLE No. 5: Determinants of Cash Conversion Cycle (CCC) in Latin American Companies

These are the result s of estimating equation (7). CCC stands for the cash conversion cycle, defined as the sum of the average collection period with theaverage period of inventory, less the average period of payment. ACCC stands for the average cash conversion cycle of the industry, considering the in-dustries detailed by the Stock Exchanges of each country. FGS represents the rate of growth of the sales for the two following years (eight quarters). PTArepresents the tangibility of the fixed assets of the company, as the company is more intensive in fixed assets this variable is closer to 1. SIZE stands forthe logarithm of sales of the company. HHIrepresents the concentration of industry and EMBI represents country risk.

Brazil Chile Mexico Peru

Type of est imation GLS GLS PWR GLS

Obs. 2738 1201 2897 2418

Number firms 77 35 88 79

Prob>chi2 0 0 0 0

Log-Likelihood -13746.64 -6514.728 -13552.12

R-sq 0.4087

ACCC 0.74*** 0.62*** 0.65*** 0.54***

p-value 0.00 0.00 0.00 0.00

FGS(t + 8) 15.33*** 14.66*** 36.18*** 7.75***

p-value 0.00 0.00 0.00 0.00

PTA -68.67*** -56.31*** -78.42***

p-value 0.00 0.00 0.00

SIZE -1.68*** -17.75*** -52.37*** -19.33***

p-value 0.00 0.00 0.00 0.00

HHI 29*** 115.4*** 148.73 61.62

p-value 0.00 0.00 0.134 0.679

EMBI -0.01*** -0.02 -0.02 -0.003

p-value 0.00 0.286 0.197 0.513

cons 39.48*** 227.08*** 684.2*** 266.12***

p-value 0.00 0.00 0.00 0.00

*** Significant at 99% level of confidence, ** Significant at 95% level of c onfidence, * Significant at 90% level of confidence

GLS: Generalized Least Squares. PWR: Prais-Winsten Regression


19/153


20/153

19

J O U R N A L

R E V I S T A

INNOVAR

19

CLASIFICACIN JEL: C45, G11, O32

RECIBIDO:febrero de 2012APROBADO:julio de 2013

CORRESPONDENCIA: ITESO, Perifrico Sur Manuel Gmez Morn 8585,Tlaquepaque, Jalisco, Mxico C.P. 45604 Tel. (33) 3669-3434. E-mail:[email protected]

CITACIN:Samaniego, A. & Mongrut, S. (2014). Relacin entre la crea-cin de valor y la inversin en I+D: una aproximacin mediante redesneuronales artificiales. Innovar, 24(51), 19-30.

Relacin entre la creacin devalor y la inversin en I+D:

una aproximacin medianteredes neuronales artificiales

ngel Samaniego AlcntarProfesor-Investigador, ITESO.Correo electrnico: [email protected]

Samuel MongrutProfesor de la Escuela de Graduados del Tecnolgico de Monterrey, Campus Quertaro,Mxico. Profesor asociado del Centro de Investigacion de la Universidad del Pacfico

(CIUP), Lima, Per.Correo electronico: [email protected]

RESUMEN: Este documento proporciona nueva evidencia sobre la interaccin de los efectos deltipo de inters a cinco aos, el nivel de apalancamiento (la relacin del pasivo y activo), las prdidashistricas, el crecimiento del PIB, la intensidad del gasto en I+D y los ingresos con el rendimientode la accin en el corto plazo. El modelo no lineal utilizado (combinacin de modelos medianteredes neuronales artificiales) explica 31.59% de la varianza del rendimiento de las acciones en losprimeros 5 meses del ao, entre 2000-2006, en los mercados AMEX, NASDAQ y NYSE. Existe evi-dencia en la literatura que el 32% de explicacin es alto, en comparacin al estudio de la literaturarealizado por Coad (2009) donde se utilizan modelos lineales que buscan explicar el crecimiento dela empresa. Se observa que los modelos lineales explican entre el 17% al 32% de la variabilidad delos datos. Al igual que en los trabajos de Bode (1998), Tsai (2005), Wang y Chien (2006) y Chien

et al. (2010) se utilizan las redes neuronales artificiales para el estudio de la interaccin entre laI+D y la empresa.

PALABRAS CLAVE:redes neuronales artificiales, inversiones en I+D, crecimiento econmico.

Introduccin

A partir del trabajo seminal de Solow (1957) existe un consenso entre aca-dmicos que gira en torno al postulado de que la inversin en investigaciny Desarrollo (I+D) incrementa la productividad y favorece el crecimientoeconmico (Griliches, 1995 y Stokey, 1995). Por otra parte, existe abun-

dante evidencia emprica de que la inversin en I+D promueve las activi-dades innovadoras en las empresas (Dosi, 1988 y Freeman & Soete, 1997).

A su vez, Irwin, Hoffman & Lamont (1998) y Hall & Bagchi-Sen (2002)verificaron que las adquisiciones de innovaciones tecnolgicas mejoran eldesempeo financiero de las empresas, lo que en la literatura se denominaI+D en productos, a diferencia de I+D en procesos.

La inversin en I+D no solo es importante por su conexin directa con elcrecimiento econmico, la productividad y el desempeo financiero empre-sarial, sino tambin porque, a pesar de lo anterior, su aplicacin a nivelmundial se encuentra concentrada en pocos pases. De acuerdo con el 2010

THE RELATIONSHIP BE TWEEN THE CR EATION OF VALUE ANDINVESTMENT IN R & D: AN APPROACH EMPLOYING ARTIFICIALNEURAL NETWORKS

ABSTRAC T:This paper provides new evidence concerning the interactionsbetween the effects of five year interest rates, leverage levels (the rela-tionship between assets and liabilities), historical losses, Gross NationalProduct (GNP) growth, levels of investment in Research and Development(R&D), income and short-term share returns. The research was carried outusing a non-linear model in which neural networks were used to combinedifferent models. The model predicted 31.59% of variation in the returnson shares traded in the AMEX, NASDAQ and NYSE markets during the firstfive months of all years between 2000 and 2006. The literature suggeststhat a 32% level of explanation is at the high end, Coads (2009) litera -ture review of models used to e xplain company growth finding that linear

models explained bet ween 17 and 32% of variability. Like Bode (1998),Tsai (2005), Wang & C hien (2006) and Chien, Wang & Lin (2010) thispaper uses artificial neural networks to examine the interaction betweenR&D and company performance.

KEY WORDS:Artificial neural networks, investment in R&D, economicgrowth.

RELATION ENTRE LA CRATION DE VALEUR ET LINVESTISSEMENTEN I + D: UNE APPROCHE LAIDE DE RSEAUX DE NEURONESARTIFICIEL S

RSUM: Ce document fournit une nouvelle vidence sur linteraction deseffets du typ e dintrts cinq ans, le niveau de levier financier (la relationentre passif et actif), les pertes historiques, la croissance du Produit Int-rieur Brut (PIB), limportance de la dpense en Investigation et Dveloppe -ment (I + D) et les rentres avec le rendement de laction court terme.Pour cette investigation, un modle non linaire a t utilis (combinaisonde modles par des rseaux de neurones artificiels) qui explique 31,59 %de la variance du rendement des actions dans les cinq premiers mois delanne, entre 2000 et 2006, sur les marchs AMRX, NASDAQ et NYSE.Il se dgage des publications sur le sujet que 32 % des e xplications estun bon niveau comparativement ltude des publications faite par Coad(2009) utilisant des modles linaires qui cherchent expliquer la crois -sance de lentreprise. On constate que les modles linaires expliquententre 17 % et 32 % de la variabilit des donnes. De mme que dans lestravaux de Bode (1998), Tsai (2005), Wang & Chien (2006) et Chien, Wang& Lin (2010), on utilise des rseaux de neurones artificiels pour ltude delinteraction entre I + D et lentreprise.

MOTS-CLS:rseaux de neurones artificiels, investissement en I + D, crois-sance conomique.

RELAO ENTRE A CRIAO DE VALOR E O INVESTIMENTO EM P+D:UMA APROXIMAO MEDIANTE REDES NEURAIS ARTIFICIAIS

RESUMO: Este documento proporciona nova evidncia sobre a interaodos efeitos do tipo de juros a cinco anos, o nvel de alavancagem (a re-lao do passivo e ativo), as perdas histricas, o crescimento do ProdutoInterno Bruto (PIB), a intensidade do gasto em Pesquisa e Desenvolvim-ento (P+D) e os ingressos com o rendimento da ao no curto prazo. Paraesta pesquisa, fez-se uso de um modelo no linear (combinao de mod-elos mediante redes neurais artificiais) que explica 31,59% da variao dorendimento das aes nos primeiros cinco meses do ano, entre 2000-2006,nos mercados AMEX, NASDAQ e NYSE. Existe evidncia na literatura deque 32% de explicao alto, em comparao com o estudo da litera-tura realizado por Coad (2009), no qual se utilizam modelos lineares quebuscam explicar o crescimento da empresa. Observa-se que os modeloslineares explicam entre 17 e 32% da variabilidade dos dados. Assim como

nos trabalhos de Bode (1998), Tsai (2005), Wang & Chien (2006) e Chien,Wang & Lin (2010), utilizam-se as redes neurais artificiais para o estudo dainterao entre a P+D e a empresa.

PALAVRAS CHAVE: redes neuronais artificiais, investimentos em P+D,crescimento econmico.



21/153



R&D Scoreboard Report1, el 78% de la inversin global enI+D ocurre nicamente en cinco pases: Estados Unidos,Japn, Alemania, Francia y el Reino Unido. Ms an, la in-tensidad de la inversin en I+D (gasto de I+D como propor-cin de las ventas) es del 3.6% (BIS, 2010).

Lo anterior es realmente paradjico debido a que se ha

demostrado que un incremento en la inversin en I+D ge-neralmente ofrece rendimientos entre el 10% y 15% o ma-yores (Hall, 1996 y Hall, Mairesse & Mohnen 2009). Esprecisamente debido a este ltimo hallazgo que el prin-cipal objetivo del presente trabajo se concentra en la iden-tificacin de los factores que influyen sobre el rendimientoaccionario de las empresas y que provocan que los inver-sionistas las compren en el corto plazo, caracterizando poraumentos o disminuciones en el gasto en I+D, as como porotros factores, a travs de un modelo no lineal.

La principal contribucin de esta investigacin es justa-

mente la utilizacin de un modelo no lineal para la pre-diccin de los rendimientos accionarios en el contexto deinversiones en I+D. La utilizacin de modelos no lineales eneste contexto ha sido sugerido por la literatura previa endiferentes contextos, por ejemplo, Tsai (2005) indic queexiste una relacin no lineal entre la productividad de lainversin en I+D y el tamao de la empresa.

Ms an, existen diversos autores que han utilizado mo-delos no lineales en el contexto de las actividades de I+D,pero desde diversas perspectivas, por ejemplo, Bode (1998)indic que es ms apropiado utilizar las arquitecturas de

las redes neuronales para la gestin de las actividades deI+D, porque tienden a ser ms simples, con menor com-plejidad y se requiere un bajo nmero de entrenamientos.

Por su parte Wang & Chien (2006) sugieren la utilizacinde una red neuronal de retropropagacin (back propaga-tion) para pronosticar los resultados de la innovacin. Losresultados de estos autores indican que esta arquitecturade red neuronal tuvo un desempeo superior a los mtodosde regresin en trminos de pronstico.

Posteriormente, Chien et al. (2010) propusieron el modeloneurofuzzy, que es una combinacin de lgica borrosa con

redes neuronales, para pronosticar los resultados de la in-novacin. Los autores encontraron que este nuevo modelosuperaba en desempeo al modelo propuesto anterior-mente, basado nicamente en redes neuronales. Por estarazn, la literatura asociada a los temas de I+D est imple-mentando cada vez ms modelos no lineales, ya que estos

1 Fue publicado por el Department for Business, Innovation andSkills (BIS), el 25 de noviembre de 2010. El R&D Scoreboard es unainvestigacin anual de las principales 1000 empresas del ReinoUnido y las principales 1000 empresas globales que invier ten en I+D.

poseen un mayor poder explicativo y permiten un mayoracierto en el pronstico.

El presente documento contiene una seccin en la que seresean las investigaciones en torno a la inversin en I+D,donde se procura evidenciar una relacin entre este aspectoy el rendimiento accionario de las empresas (mediante el

denominado estudio de eventos), por una parte, y el valorde la empresa (mediante modelos lineales), por otra. En lasiguiente seccin se resean investigaciones emblemticasque utilizan las Redes Neuronales Artificiales (RNA) en elcontexto financiero, ya que -segn nuestro conocimiento-no existen trabajos que las utilicen simultneamente enel contexto financiero y con inversin de I+D. En las dossiguientes secciones, se describe la data utilizada y se pre-sentan los modelos lineales y no lineales que sern compa-rados en la estimacin, respectivamente. Para terminar, sediscuten los resultados y se concluye el trabajo.

Estudio de la literatura

En la literatura se encuentran varios trabajos donde la in-novacin se evidencia como una seal de oportunidadesde crecimiento empresarial, aunque no est muy claro qufactores externos al anuncio propician la credibilidad delinversionista antes de que los proyectos asociados a estosanuncios obtengan resultados reales. Como se coment,las inversiones en I+D juegan un papel clave en la innova-cin empresarial y su anuncio puede provocar que el inver-

sionista burstil invierta en este tipo de empresas.

Para el mediano plazo, Chan et al. (2001) examinaron laintensidad del I+D (en valor) y observaron que el mercadoes pesimista cuando empresas de alta tecnologa dismi-nuyen su gasto en esta rea. Adems este tipo de em-presas posee una alta relacin entre el gasto en I+D y elvalor de sus activos (a valor de mercado) y experimentanaltos rendimientos futuros (3 aos).

En largo plazo, Eberhart, Maxwell & Siddique (2004) exa-minaron eventos mayores al 5% de incremento en la in-

tensidad de I+D (en cambios porcentuales), encontrandouna baja reaccin del inversionista ante este evento en unplazo de 5 aos.

McNamara & Baden-Fuller (2007) estudiaron el efecto enlos rendimientos accionarios producidos por 1277 anunciosde gastos en I+D, dados por 178 empresas farmacuticasen tres etapas del desarrollo en I+D: pruebas pre-clnicas,pruebas clnicas en seres humanos y explotacin comercial.Los autores encontraron rendimientos anormales entre 3 y4%, siendo superiores en la etapa inicial para empresas de


22/153

J O U R N A L

R E V I S T A

INNOVAR


menor tamao e iguales en todas las etapas para empresasde gran tamao.

Moltchanski, Nagm & Powell (2010) estudiaron la res-puesta del mercado de capitales australiano, durante elperiodo 1994-2006, ante anuncios de gastos de I+D, anun-cios de gastos de capital (CAPEX) y anuncios de gastos enTecnologa de la Informacin (TI). Mediante un estudio deeventos encontraron rendimientos anormales acumuladosentre 3 y 5%, asociados a los gastos en I+D y TI.

Samaniego, Trejo-Pech, Mongrut y Fuenzalida (2011) en-contraron evidencia de rendimientos anormales positivosen el corto plazo (9 das) ante incrementos en la inten-sidad del I+D (cambios en valor absoluto), pero no evi-

dencia estadsticamente significativa de rendimientosanormales positivos ante decrementos en le intensidad deI+D. Adems, para empresas clasificadas como de alta tec-nologa la reaccin era mayor.

Alternativamente, otros autores decidieron estudiar la re-lacin entre el valor de la empresa u otra variable aproxi-mada con las inversiones (o gastos) en I+D, haciendo usode otras variables como tamao y concentracin de pro-piedad. Lo interesante es que todos los estudios han utili-zado modelos li

persistencia de la rentabildad chile ppg114

Documents