Firms' Innovation, Public Financial Support, and Total Factor Productivity: The Case of Manufactures in Peru

• DOI: http://doi.org/10.1111/rode.12147
• Published date: 01 May 2015
• Date: 01 May 2015
• Author: Mario D. Tello

Firms’ Innovation, Public Financial Support, and

Total Factor Productivity: The Case of Manufactures

in Peru

Mario D. Tello*

Abstract

Based upon an adjusted Crepon–Duguet–Mairesse (CDM) model, this paper analyzes the relationship

between investment intensity, public financial support, innovation, and total factor productivity (TFP) for a

sample of manufacturing firms of Peru with data obtained from the 2004 survey of science, technology, and

innovation (STI) activities. The estimation of the model indicates that large firms are more likely to invest

in STI activities and firms’ size increases the probability of producing technological inovation (TI) and non-

technological innovation (NTI). STI firms’ investment intensity and public financial support have also

helped manufacturing firms to increase the probability of producing TI outcomes. Further, such support

may have increased firms’ investment on STI activities. The innovation effects on TFP, however, were sta-

tistically not clear or robust. Thus, whereas investment intensity did increase firms’ TPF in low-tech manu-

facturing firms, this is not the case for high-tech firms. For this group of firms, relatively high capital–labor

ratio and the availability of a high level of human capital seem to promote higher levels of TFP.

1. Introduction

This paper analyzes empirically the relation between STI investment, innovation, and

total factor productivity (TFP) at the firm level for a sample of Peruvian manufactur-

ing enterprises. The low contribution of TFP to economic growth in the first decade of

this century for Latin American Countries (LACs) has been documented in several

studies (e.g. Pagés, 2010; Astorga et al., 2011; Ferreira et al., 2013). Furthermore, in

the same decade R&D expenditures out of GDP, a key determinant of innovation has

been very low relative to developed countries (World Economic Forum (WEF),

1998–2012). Absence of data at the firm level, however, has limited the analysis of the

relationship between innovation and TFP in several ‘innovation shortfall’ countries1

as affirmed by Demmel et al. (2013). In this sense, a key contribution of this paper is

to use and merge two sets of firms’ data from the Peruvian manufacturing sector to

address such a relationship.

The empirical literature focusing on the relationship between innovation and prod-

uctivity (e.g. Hall, 2011; OECD (Organisation for Economic Co-operation and

Development), 2009; Álvarez and Crespi, 2011; Steingraber and Gonçalves, 2011) is

ample and has been originated by the availability of data from manufacturing surveys

* Tello: Departamento de Economía de la Pontificia Universidad Católica del Perú, Av. Universitaria 1801,

San Miguel, Lima 32, Perú. E-mail: mtello@pucp.edu.pe. This paper is a modified and shortened version of

the paper presented at the 18th DEGIT conference held in Lima, Peru, in September of 2013 and was fin-

ished when the author was a visiting CAF fellow of the Latin American Centre at the University of Oxford.

Also the paper is the result of the project on innovation and productivity in Latin America financed by

IADB, CINVE, and IDRC. The author thanks the participants of the DEGIT conference and project semi-

nars, particularly Gustavo Crespi, Fernando Vargas, Diego Aboal, and Sylvia Dohnert, and two anony-

mous referees of this journal. In addition, the author thanks Carla Solis, Mayte Ysique, and Paulo Matos

for their excellent research assistant work.

Review of Development Economics, 19(2), 358–374, 2015

DOI:10.1111/rode.12147

© 2015 John Wiley & Sons Ltd

at the firm level in developed and a few developing countries. Since the pioneering

studies of Griliches (1979) and Griliches and Pakes (1980, 1984), this literature has

been developed under the dominant Crepon–Duguet–Miresse (CDM) framework

(see Crepon et al., 1998) and occasionally for specific micro models (e.g. Guadalupe

et al., 2012). Whereas CDM econometric models postulate a system of ad-hoc equa-

tions that links, firms’ R&D investment decision, investment intensity, innovation,

and productivity, specific micro models are based upon structural models of the

process of innovation of firms.

Given the empirical fit of CDM econometric models in past studies and their sound

theoretical base (since it relates to key aspects of the process of innovation of firms),

such models seem to be a reasonable starting point for the analysis of Peruvian data.

Consequently, based upon an adjusted CDM model, the main contribution of this

paper is to present evidence of the relationship between STI investment, innovation,

and productivity for a sample of manufacturing firms in Peru using data from two

surveys on the innovation of firms (CONCYTEC–INEI, 2005) and on the productive

activities of firms (the INEI, 2002–2007). This adjusted CDM model allows estimating

the role of public financial support on firms’ decisions to invest in STI activities, the

amount or intensity of such investment, the TI and NTI outcomes, and indirectly on

the TFP of manufacturing firms in Peru.

There are two key distinctions between this paper and previous studies, particularly

that of Demmel et al. (2013). First, it takes estimates of the TFP derived from an

Olley–Pakes (1996) estimation of a Cobb–Douglas production function from a panel

sample of manufacturing firms in Peru for the period 2002–2007. The study of

Demmel et al. (2013) estimate TFP using a cross-section sample of manufacturing

firms from four countries (Argentina, Mexico, Colombia, and Peru). Second, it uses a

system of five equations that relates firms’ decisions to invest in STI activities, their

investment intensity, and innovation outcomes, and their effects upon TFP. Demmel

et al. (2013), in contrast, uses one equation given the data limitation (World Bank,

2010)’. The rest of the paper is organized in four sections. Section 2 presents a brief

summary of the literature. Section 3 describes the CDM model. Section 4 presents the

estimation strategy and reports the econometric results. The final section (section 5)

summarizes the main findings.

2. Literature Review

CDM models have two main features: on the one hand, they define an ad-hoc struc-

tural model in which the main variables of the process of innovation of firms are

related. On the other hand, econometric techniques are used to deal with selectivity,

simultaneity biases, and some statistical features of the available data. These models

consist of four stages: (i) a firm’s decision to invest in R&D activities—this is the

firm’s R&D investment decision equation; (ii) a firm’s decision on the amount to

invest—this is the firm’s research intensity equation; (iii) the production of knowledge

(technology) derived from this investment (the “knowledge production” function, e.g.

Griliches, 1979; Griliches and Pakes, 1984)—this is the firms’ innovation output equa-

tion; and (iv) output is produced using new knowledge (technological innovation)

along with other inputs—this is a firm’s productivity equation (Crespi and Zuñiga,

2010). In addition to firms’ characteristics, CDM models may incorporate some exter-

nal forces acting concurrently on the innovation decisions of firms. These are tradi-

tional indicators for demand-driven innovation (i.e. environmental, health, and safety

THE CASE OF MANUFACTURES IN PERU 359

© 2015 John Wiley & Sons Ltd