3.

Investing in Ideas: Business Innovation Policies

[4 boxes, 9 figures, 7 tables, 12,740 words]

Since the pioneering work of Solow (1957), technological change has been credited with

explaining a substantial share of economic growth. Indeed, recent evidence for the United States

shows that investments in research and development (R&D)—a proxy for the innovation effort

of a nation—made up 40 percent of the productivity growth observed during the postwar era

(Reikard, 2011). Based on these findings, several Latin American and Caribbean countries have

established and implemented public policies aimed at enhancing innovation. In practice, the first

explicit interventions to encourage innovation by the private sector emerged even earlier, toward

the end of World War II. Although many of these policies were either abandoned or dramatically

downsized under the structural reforms inspired by the Washington Consensus, the disappointing

results in terms of productivity growth have led several countries in the region to reintroduce

policies to stimulate innovation and encourage technology adoption.

Since the early 1990s, a new generation of public programs to encourage business

innovation has spread throughout the region. Policy experimentation is already generating new

evidence about the effectiveness of these interventions. This chapter assesses the main

developments in innovation policies, taking stock of the learning achieved so far.

A Global Perspective: Latin America Lags Behind

Assessing innovation is a very challenging task. Innovation can be defined as the

implementation of new or significantly improved products, processes, services, or organizational

models (OECD, 2005b). As such, innovation is more than a technological improvement; it

includes changes in organizational and marketing models. Innovation encompasses minor

changes as well as breakthroughs. Clearly, innovation is a highly subjective concept, as each

person may have a very different understanding of how novel a device or a process is. For this

reason, comparisons of innovation indicators are colored by individual biases.

However, innovation can be approximated by measuring certain “aspects” of the

innovation process. Griliches (1979) suggests that similar to the production of goods, the

production of ideas can be explained by a knowledge production function, where innovation

results from firm investments in R&D and the stock of knowledge. More recent innovation

research has expanded the set of inputs to include human capital, training, machinery, licenses,

software, and the like. Thus, it is possible to garner a glimpse of innovation by measuring some

of those inputs. On the outputs side, approximations for innovation outcomes include

productivity indexes, numbers of intellectual property rights, scientific publications, and self-

reported figures of innovations collected from innovation surveys (Smith, 2006). Each one of

these indicators by itself provides a partial view of the innovation process; however, together

they offer a reasonably comprehensive picture of the innovation process of a firm.

Figure 3.1 assesses the performance of Latin America and the Caribbean vis-à-vis

developed countries with regard to innovation inputs. Panel (a) summarizes aggregate R&D

intensities for a sample of Latin American and Caribbean countries and compares them with a

sample of developed countries. This panel highlights several issues. First, R&D intensities in
Latin America and the Caribbean are systematically lower than in developed countries. Second,

according to this indicator, the world’s top performers are precisely those countries that have

managed to catch up with other developed countries over the last 20 or 30 years: Israel (4.3

percent), Finland (3.9 percent), and South Korea (3.7 percent). Third, in top-performing

countries, the private sector finances a large proportion of the R&D effort. While in developed

countries firms explain more than 60 percent of the national investment in R&D, this figure is

less than 35 percent in Latin America and the Caribbean. These findings suggest an important

deficit in R&D investment in the region, particularly in the private sector.

Panel (b) compares average investment intensities in innovation as a percentage of sales,

including not only R&D but also training, purchasing of machinery and equipment, software

licenses and royalties for the use of patented technology for a sample of countries using

information from innovation surveys. Even within this broader definition of innovation

investment, a significant gap exists. While the average firm in a developed country spends

almost 4 percent of sales on innovation, the typical firm in Latin America and the Caribbean

spends around 2.5 percent. The gap is particularly large in intangible investments such as R&D.

The pattern that emerges is that technology embodied in machinery, mostly imported from

abroad, is the main force of innovation investment in the region. The evidence from developed

countries suggests that relying on imported technology is not necessarily bad if it leads to

domestic learning. However, in order for this to happen, technology must be combined with

absorptive capacities that allow for further improvements. Absorptive capacities depend on

research and development efforts and complementary human capital. According to panel (c),

there are on average only 1.1 researchers per 1000 workers in the region, eight times fewer than
in the typical OECD country—even though the average number of researchers in the region

increased by 50 percent from 2000 to 2010. 1

[Insert Figure 3.1 Innovation in Latin America and the Caribbean at a

Glance]

In summary, the Latin American and Caribbean innovation process is based on the

adoption and incremental improvement of existing technologies, rather than investment in R&D.

Has this pattern allowed the region to catch up to the rest of the world in terms of productivity?

The results in panel (d), which shows average productivity growth rates for each county from

1960 to 2010, suggest otherwise. In fact, long-term productivity growth rates in Latin American

and Caribbean countries are systematically lower than those in the OECD. Moreover, having a

productivity growth higher than the United States—depicted by the vertical line in panel (d)—is

the exception in Latin America.

The Rationale for Public Policy

Innovation is the result in large part of investment decisions made by firms; these

decisions are affected by the same conditions that affect investment in general. Indeed, the

quality of regulation, protection of property rights, tax code, macroeconomic regime, the

intensity of competition, and infrastructure development all affect investment decisions on

innovation—sometimes even more significantly than for fixed capital investments (OECD,
2013a). However, having the right framework conditions is a necessary but not sufficient

condition for innovation. Since most countries in the region have internalized the importance of

these conditions and made important progress with them, the focus of this chapter is on explicit

innovation policy measures that still need to be implemented.

Although it is generally true that every modern economy needs an innovation policy, a

flaw in the design of innovation policy tends to be its focus on symptoms rather than on the

actual, underlying causes of underinvestment. Typically, innovation policy is justified on the

basis of gaps in R&D or technology adoption when compared with benchmark economies. This

focus on symptoms rather than constraints normally leads to poor policy design: low investment

in innovation or low levels of technology adoption could also be an optimum response to low

returns. In other cases, government justifies innovation policy based on socially desirable

objectives, such as job creation or social inclusion, without realizing that the relationship

between these objectives and innovation is complex and largely ambiguous.

Theory-based Justifications

The fundamental premise for innovation policies is that government intervention can be

beneficial if profit-driven actors underinvest from a social welfare perspective (Steinmueller,

2010). Broadly speaking, the rationale for public policy in this field can be based on the

following considerations:
Spillovers and the “Public Good” Nature of Knowledge

Since the seminal works by Nelson (1959) and Arrow (1962), knowledge has been

regarded as a nonrival 2 and nonexcludable 3 good. If knowledge does indeed have these

properties, then a firm’s rivals may be able to free-ride on its investments. These spillovers may

create a wedge between private and social returns and a disincentive against private investment

in knowledge production. However, spillovers are not automatic and should not be taken for

granted in every circumstance, as not all knowledge enjoys the properties of a public good with

the same intensity. Certainly, the “public good” rationale of knowledge applies more strongly to

generic or scientific knowledge than technological knowledge, which is more applicable and

specific to the firm. 4 Furthermore, in order for the public good rationale to be valid, there should

be some possibility of free-riding. If the originator can protect the results of the knowledge

generated (through entry barriers or the use of strategic mechanisms, for example), then the

potential for market failure declines. On the other hand, knowledge generated through

collaboration among different parties might be more difficult to protect and therefore more prone

to spillovers than knowledge generated by individual entities.

The Problem of Asymmetric Information and Uncertainty

Innovation projects distinguish themselves from ordinary investments in several ways

(Hall and Lerner, 2010). First, the returns to innovation investments are more uncertain and

involve longer gestation lags. Second, innovators may be reluctant to disclose detailed

information about their projects because of spillovers. Third, innovation investments normally
include a large proportion of intangible assets (such as human capital) that have very limited use

as collateral. Although the problem of asymmetric information is always present whenever the

investor and financier are different entities, this problem may be worse in the case of knowledge

investments. This creates a wedge between the rate of return required by an innovator investing

his or her own funds and that required by external investors. Unless the innovator is particularly

wealthy, privately (and maybe socially) profitable innovation projects may not materialize due to

lack of access to financing or to the high cost of capital.

There may also be asymmetric information with respect to the knowledge about available

technologies. The most traditional diffusion model—in which technology adoption results from

the spread of information about the technology—highlights the fact that diffusion is not

automatic. In a world of imperfect information, policy intervention would therefore focus on

providing information, such as demonstration projects, advertising campaigns, technology

monitoring exercises, and extension services that inform an industry of recent technology

advances.

The Pervasiveness of Coordination Failures

Knowledge also has important tacit components that cannot be embodied in a set of

artifacts, such as machines, manuals, or blueprints. Thus, firms can benefit from networking with

one another and other actors because they need to learn from the knowledge bases of other

organizations. However, coordination failures can hinder the effectiveness of these knowledge

networks. Coordination failures emerge whenever private and public agents fail to coordinate

their knowledge investment plans in order to create mutual positive externalities (Aghion, David,

and Foray, 2009). Coordination failures also emerge in the process of accessing technological
infrastructure. Firms that alone cannot afford infrastructure can gain access to it if they

collaborate with others. Solving coordination problems requires paying special attention to those

institutional failures that can affect the linkages between the different actors in the innovation

system.

It is argued that one of the few advantages of a developing country is that it can simply

free-ride on the innovation investments of developed countries. As is clear from the preceding

discussion, the real world is far more complex than this. The returns of a given technology

depend on the context in which it is used. Key complementary inputs, such as human capital,

institutions, and natural resources, may vary greatly across different locations and affect the

performance of the same technology in different places. In order to successfully adopt a given

technology, firms must discover whether this technology is suitable for each particular context.

For that, local investment in learning and innovation is needed. These investments are affected

by the same problems of spillovers, asymmetric information, and coordination that affect

innovation investments in general. To complicate matters, in developing countries many of these

market failures coexist and feedback into each other. So, unfortunately for developing countries,

there is no free supermarket of ready-to-use ideas.

Reassessing the Innovation Investment Gap

The evidence so far suggests that Latin America and the Caribbean seriously under-

invests in innovation. However, this statement lumps together countries that are very different.

The observation that an investment gap in intangibles exists is not enough to suggest that an

economy faces an innovation problem. Low investment in R&D can also reflect systemic

problems that affect the accumulation of all sorts of assets, including physical and human capital
(Maloney and Rodríguez-Clare, 2007). A different explanation for an R&D intensity gap is

production specialization. The propensity of firms to invest in R&D varies across sectors (Pavitt,

1984). In an open economy, specialization depends on factor endowments, preferences, and

relative productivities across sectors, clearly affecting the validity of the country comparison of

aggregate indicators. 5

Several mechanisms might affect firms’ decisions to invest in innovation activities,

mostly through their access to generic knowledge, human capital, and finance. The degree to

which a society develops institutional arrangements to consistently provide these three

complementary inputs affects private investment in innovation. Table 3.1 shows how much these

variables explain the gap in innovation investment between Latin America and the Caribbean

and OECD countries. In the period 1995–2010, the business sector of the typical Latin American

and Caribbean country invested in R&D 1.18 percent of GDP less than the typical OECD

country, and this gap has increased since the period 1980–94, when it was 0.90 percent of GDP.

The factors underlying these gaps have changed. Prior to 1995, low access to generic

knowledge explained about 30 percent of the gap, while the absence of dynamic sectors in the

production structure explained just 10 percent. The lack of human capital and financial

development were in between those figures. After 1995, however, human capital increased

significantly in importance and, more strikingly, the role of the production structure increased

markedly to explain 26 percent of business investment gaps. On the other hand, financial

development, and to a lesser extent generic knowledge, have become less important. Clearly, the

Latin American and Caribbean business sector suffers from an innovation investment shortfall

beyond what would be expected given the financial development and human capital

accumulation of the region. Moreover, low public sector investment in the generation of generic
knowledge and the lack of sophistication of the production structure explain a significant part of

this gap. 6

[Insert Table 3.1 Explaining the Gaps in Business R&D: The OECD vs.

Latin America and the Caribbean (percent of the total GAP)]

Weighing the Social Returns

The evidence of investment gaps is not enough to justify intervention because they could

reflect a lack of innovation opportunities. Assessing this possibility requires looking at the social

returns rates (SRRs) to innovation investments. A review of more than 50 years of research

suggests that social returns to R&D are strongly positive (Hall, Mairesse, and Mohnen, 2010).

However, much of this research focuses on evidence from developed countries. Lederman and

Maloney (2003) find that the returns to R&D are not only higher for developing countries, but

also higher than the estimated return on physical capital. More specifically, for Latin American

and Caribbean countries, Maloney and Rodríguez-Clare (2007) find that social rates of return

calibrated using international data vary between 51 percent in the case of Peru and 16 percent in

the case of El Salvador, with a regional average of 33 percent.

To what extent do investments in R&D contribute to productivity growth? Griffith,

Redding, and Van Reenen (2004) propose an approach in which productivity growth is the result

of innovation and technology transfer, and investments in R&D not only stimulate innovation but

also build absorptive capacities for technology transfer. The social return rates to R&D are the

combined results of these two forces.

 In order to implement this methodology, the Pagés (2010) productivity data set is

expanded to include data on R&D investments from different sources. Figure 3.2 summarizes the

results of this method and suggests that SRRs for R&D investments are not only systematically

higher in Latin America and the Caribbean than in the OECD (56 percent vs. 32 percent in 2007)

but also that they have followed a divergent path over time. Indeed, while social returns have

declined in the OECD (mostly due to the lower returns from technology transfer as these

countries have moved closer to the technological frontier) in Latin America and the Caribbean,

social returns to R&D have tended to increase (mostly due to the greater scope for technology

transfer as these countries have systematically shifted away from the technological frontier). This

increasing trend in social returns, together with growing investment gaps, are consistent with the

finding that the Latin American and Caribbean region indeed faces an innovation shortfall.

[Insert Figure 3.2 Social Returns to R&D, Latin America and the

Caribbean vs. the OECD]

What Works in Innovation Public Policy

Innovation policy covers a broad set of issues that have been in the policy agenda for

decades. Although countries vary widely, the experience of successful catch-up economies

suggests several key commonalities (OECD, 2005a).

• A long-term, public-private consensus exists on the importance of maintaining public

support and continuously updating innovation policies. Indeed, many of the best
 performers not only boast high innovation investment rates today, but also sustained high

rates of effort over long periods of time, even above what was expected given their GDP

per capita (see Figure 3.3).

• An early and strong focus on investment in adoption of foreign technology, research

infrastructure and human capital 7 is coupled with support to applied research in key

sectors or technologies, 8 and rather weak IPR protection.

• A continuous effort to improve framework conditions (macro stabilization, trade

openness, fiscal balance, competition, regulation, IPR protection, and the like) echoes

measures recommended by the Washington Consensus.

• Reforms centered on framework conditions were simultaneously accompanied by

increasing support for investments in science and technology and business innovation,

allowing for a continuous shift of resources toward the most dynamic sectors.

• Policy instruments such as financial entitlements for public technological institutes,

grants, and tax incentives for business innovation, mission-oriented research funding and

public procurement were used extensively, and refocused continually on the generation of

spillovers (research collaboration and technologies that spread across different sectors,

such as biotechnology and information and communication technology) (see Box 3.1).

• A continuous focus on institutional capacity building, monitoring, and evaluation fed into

policy learning and gradually fostered a more complex and focused policy mix. South

Korea illustrates a typical example of this process of policy building (see Box 3.2).

[Insert Figure 3.3 Star Performers' R&D Investment]

 The current innovation policy landscape in developed countries encompasses many

designs. Based on the framework proposed in Chapter 2, these multiple policy designs can be

organized along two dimensions: scope, depending on whether the policy focuses on the

provision of public inputs or market interventions; and type, depending on whether it is

horizontal (economy-wide) or vertical (sector-focused). These two dimensions define a 2x2

matrix that reclassifies innovation policies in four quadrants. Table 3.2 presents several examples

of scope-type combinations of policy instruments that are commonly found in successful catch-

up economies.

[Insert Box 3.1 Fostering Innovation through Government-sponsored

Mission-oriented Research and Public Procurement: The Case of the United

States]

[Insert Box 3.2 Innovation Policy Building through Catch-up: The Case of

South Korea]

[Insert Table 3.2 Innovation Policies in Developed Countries: A Taxonomy]

The Latin American Experience with Innovation Policy

The Latin American experience with innovation policy dates backs to the 1950s and since

then different policy approaches have been tried and abandoned over time. Although country

experiences are very idiosyncratic, three broad policy paradigms have been tried: the supply side

approach, which extended from the 1950s to the 1980s; the demand side approach, which
reigned from the 1980s through the 2000s; and the (emerging) systemic approach, which has

increasingly become the focus of policy interest since the 2000s.

The supply side approach (1950s–1980s) was based on the idea of linearity from supply

to demand: direct production of knowledge and complementary assets—in particular, human

capital and information—by public institutions (such as laboratories, research institutes, and

universities, mostly funded through entitlements) gave way to a series of instruments in the

public inputs/horizontal quadrant of Table 3.2. New institutions—the national research

councils—governed the system and were tasked with funding research, supporting human capital

formation, and establishing policy frameworks. Technological institutes were established to

complement support to research and technical and professional training. 9 Operating at the sector

level, these institutes fulfilled a dual role: carry out applied research, and transfer knowledge to

firms operating in strategic sectors. Private sector development of technological absorptive

capacities and linkages between knowledge supply and demand were given far less importance.

The demand side approach (1980s–2000s) featured the structural reforms of the

Washington Consensus. The diagnosis blamed government failures and argued that keeping

intervention to a minimum was the best way to avoid them. This phase had important

implications for innovation policy. The majority of the public organizations and institutes

designed to promote innovation lost importance within the state bureaucracy. Public budgets

were severely curtailed. New incentive regimes were set up to introduce market discipline in

technological institutes that had to increase their funding by selling services to the private sector.

Human capital formation was deregulated and private universities entered the market while

intellectual property frameworks were gradually strengthened. All these changes occurred as

productivity growth at the technological frontier accelerated dramatically (Sagasti, 2011).

Ironically, just as incentives to innovation on the demand side were enhanced (mostly through

product market competition), support to the supply side of the equation was dismantled.

The pitfalls of the reforms became evident toward the end of this phase, when it became

clear that spillovers, lack of complementary assets, and financing were important obstacles for

firms to adapt to the new scenario. In response, some countries began experimenting with new

horizontal/market policy instruments, introducing grants for business R&D, R&D tax credits,

conditional loans, and vouchers for technology transfer in the second half of the 1990s. Most of

these programs were delivered through technology development funds, which initially worked

out of existing institutions such as development agencies or research councils, and later spun off

into dedicated agencies or funding units.

The systemic approach, which began in the early 2000s, emerged from a growing

consensus that business innovation support with a strong focus on the individual firm was not

enough to internalize spillovers and solve coordination failures. The diffusion of the idea of an

innovation system triggered renewed interest in investing on the supply side, but with increased

concern about generating the right incentives to favor closer coordination between supply and

demand. New institutions, such as technology liaison offices specialized in linking the different

actors, emerged. After many years of inaction, there was renewed interest in supporting

technology extension, but now with a focus on building absorptive capabilities in small and

medium enterprises (SMEs). Dissatisfaction with purely horizontal policies also grew. Thus,

since the early 2000s, the impetus has shifted to vertical programs. Some countries are

experimenting with funding schemes in areas where public procurement is important (such as

programs in health and energy) and are targeting subsidies to technologies (such as information

technologies) that can spread out across the production sector at large. This proliferation of
programs with very different designs and implementing agencies has heightened the stress on

institutions and highlights the need to improve policy coordination. Currently, the mix of

innovation policies in some countries is not very different from that in developed economies

depicted in Table 3.2.

The Innovation Tool Box

The previous analysis offers different justifications for the implementation of innovation

policies based on the idea that profit-seeking agents will produce both a level and direction of

knowledge that fall short from a social welfare perspective. As David, Hall, and Toole (2000)

note, public policy has followed two main approaches: direct production of knowledge by public

institutions (universities, laboratories and public research institutes), most of them under the

category of public goods interventions; and fiscal incentives for private investment in knowledge

generation. 10

Issues of governance and funding incentives of public technological institutes are beyond

the scope of this chapter; in any case, their public funding is normally far less controversial than

market interventions. This section focuses on the second class of policy interventions, in

particular, two types of incentives: fiscal incentives for innovation, whose goal is to increase

private sector investment in innovation and whose focus is the firm as a knowledge producer;

and technology extension programs, whose goal is to stimulate firms to acquire or improve their

use of technology and whose focus is the firm as a knowledge user. Fiscal incentives mostly

occur in the form of direct subsidies or tax incentives, while technology extension programs

cover a wider set of interventions. Although other instruments exist, this discussion focuses on
these two because they are the most prone to moral hazard and rent-seeking problems, as they

imply a net transfer of resources to the private sector.

Fiscal Incentives

Fiscal incentives come in two different forms: subsidies and tax incentives. Subsidies are

a type of direct support for business innovation that is project-specific. They transfer cash to

firms on the condition that firms execute a series of innovation activities. Subsidies are normally

delivered through two different mechanisms: nonreimbursable grants and conditional loans. 11 By

contrast, tax incentives are based on firm-level innovation activities and operate through a

reduction of the firm tax liability.

Fiscal incentives not only reduce the marginal cost of capital to undertake innovation

investments, but can also encourage collaboration with other actors in the innovation system,

such as research centers, technological institutes, or firms. Table 3.3 summarizes the main

features of both types of fiscal incentives.

[Insert Table 3.3 Fiscal Incentives: Subsidies versus Tax Incentives]

Two important differences between subsidies and tax incentives relate to their relative

effectiveness to internalize spillovers and their focus on those segments of the population of

firms most likely to be affected by market failures. Subsidies, given their project-based nature,

can not only target high spillover projects but can also direct funding toward firms more severely
affected by market failures, such as young firms and innovative SMEs. These features are less

clear in the case of tax incentives. In the first place, tax incentives are a fully market-friendly

mechanism; the firm decides which projects will be implemented. This might bias the incentives

toward projects with higher private appropriability. Moreover, tax incentives are not proportional

to the difference between social and private rates of return, as they should be, but instead to total

R&D costs (Maloney and Perry, 2005). Tax incentives also suffer from poor targeting as their

impacts depend on the fiscal position of each firm—an aspect that normally biases them toward

large firms. 12 In the case of subsidies, the fiscal cost is under control, since the resources

allocated to the programs are budgeted. On the other hand, in the case of tax incentives, fiscal

costs are less certain because the fiscal authority has no control over firms’ decisions. Both types

of fiscal incentives suffer from moral hazard problems; however, several good practice designs

are available to mitigate them (see the bottom panel of Table 3.3). Implementing these good

practices requires capacity building in both innovation agencies and tax authorities. The higher

administration and compliance costs of the direct subsidies must be compared with the higher

policing costs of the tax incentive. In summary, subsidies have a series of nice design features

that might make them a preferred choice compared to tax incentives for several Latin American

and Caribbean countries, in particular those with very weak tax authorities.

Sometimes subsidies coexist with credit programs that focus on funding the adoption of

innovative technologies by firms (particularly technologies embodied in machinery and

equipment). The rationale for the credit lines differs from that normally used for subsidies. In the

case of the adoption of technology embodied in machinery and equipment, the returns are more

likely to be appropriated by the investors, so policy should focus on solving a liquidity constraint

problem only through the provision of credit. However, if the purchased technology generates
spillovers to the rest of the sector—for example, through demonstration effects by the pioneer

adopting firm—the credit could be combined with a subsidy to encourage technology

dissemination. The implementing agency must carefully define what should or should not be

considered an innovative technology and the potential for spillovers (see Box 3.3).

[Insert Box 3.3 FONTAR’s Toolkit: Basic Rules for Allocating Subsidies

vs. Credit]

So far in the region, following the trends from developed countries, subsidies have

increasingly focused on fostering research collaboration and linkages among the different actors

in the innovation system. Indeed, countries are gradually moving from supporting single projects

that involve university-industry interaction to more integral programs that involve full sectors by

making use of technological consortia. A technological consortium seeks to pool innovation

efforts of different firms and to increase their collaboration with universities and research

centers, which has been a historical problem in Latin America and the Caribbean. The basic

assumption is that these consortia could help internalize spillovers, coordinate complementary

private investments and reduce duplication of competitive research agendas. By solving these

multiple failures, consortia should trigger higher returns to R&D investment. 13 Although still too

early to make a final judgment on the performance results of these collaborative interventions,

early evidence suggests that the results critically depend on correctly aligning the incentives of

firms and universities, the absorptive capacities of participating firms, the proximity among

participants and the experience and conflict solving capabilities by the group’s manager. On the

public sector side, these large-scale interventions also demand coordination among different
institutions and flexibility as the research agenda emerges from the interaction among the

different actors (Álvarez, Crespi, and Cuevas, 2012)

Figure 3.4 summarizes the amount of public resources spent on business innovation for

the countries with the largest support systems in the region. Brazil tops the ranking, with

transfers to the private sector amounting to 0.14 percent of GDP in 2008. About 70 percent of

these resources were delivered through subsidies, and 30 percent through tax incentives. Chile

and Colombia came next (0.04 percent of GDP). Figure 3.4 also shows that the typical OECD

country spends 0.11 percent of GDP on R&D fiscal incentives, around four times more than the

typical Latin American and Caribbean country. The experience of developed countries suggests

that scaling up the system strongly depends on the generosity of the tax incentive component of

the policy (the same is true in Latin America and the Caribbean).

[Insert Figure 3.4 Direct Government Funding and Tax Incentives for

Business Innovation (percent GDP)]

Technology Extension

The second type of market intervention to promote private innovation is a technology

extension program (TEP). TEPs are designed to facilitate the adoption of existing technologies to

improve the efficiency of a production unit. What differentiates a TEP from other innovation

policy tools is that the new technology is mostly developed outside the adopting unit. 14 These

programs typically have a dual focus. On the one hand, they provide services to reduce the costs
of searching for information about new technologies, sometimes matching user needs with

appropriate suppliers. They also provide support to enhance firms’ ability to absorb new

technologies, through hands-on training, pilot production demonstration, and assistance when

negotiating with the technology supplier (De Ferranti et al., 2003). In some cases, TEPs are

combined with support for incremental innovation (in terms of adaptation to local conditions).

As in the case for innovation, cofinancing is a key component of TEPs.

Different models and approaches to TEPs have been developed and implemented over

time. 15 The approach in the United States and the United Kingdom—spearheaded by the U.S.

Manufacturing Extension Partnership (MEP) and England’s Manufacturing Advisory Service

(MAS)—has traditionally focused on intervening at the firm level by providing field services to

enhance the ability of SMEs to adopt new technologies or organizational models. The continental

European approach, typified by Austria, Germany, and some Scandinavian countries, focuses

more on supporting not only the SME’s adoption of existing external technologies, but also on

improving these technologies through sector-specific research consortia. The Japanese and

Canadian approaches borrow elements from both.

Latin American and Caribbean countries have traditionally offered technological

extension services to SMEs through national technology institutes (NTIs), both in manufacturing

and agriculture, some of them established during the early 1950s. Problems related to funding

and incentives have compromised the effectiveness of these institutions. Most of the NTIs have

been reformed, and funding through entitlements is gradually being replaced with performance

agreements and sales of services. Moreover, the traditional NTI model has been complemented

with various programs aimed at creating a market for technological services. Most Latin

American and Caribbean governments have introduced programs based on public-private cost
sharing, sometimes with vouchers for producers to purchase services from publicly certified

private providers and NTIs. These changes have gradually led to the generation of a private

supply of technological services for SMEs. Finally, TEPs in Latin America and the Caribbean

have progressively moved from targeting individual firms to groups of firms. This change

responds to the importance of sharing complementary skills and of accessing certain

technological infrastructure. To help overcome potential coordination failures, new TEPs have

been developed to promote firm collaboration and networking. Although assessments suggest

that these changes are steps in the right direction, as TEPs have effectively improved the

performance of their target population, they have also revealed structural limitations in reaching

firms that suffer severe financial and technological constraints. For this reason, fully publicly

operated services may still be required to satisfy demand for extension services at the base of

productivity pyramid. Another limitation of TEPs in the region is that despite the reforms,

coverage—in terms of the proportion of assisted firms over target population—seems to be very

low (Figure 3.5).

[Insert Figure 3.5, Manufacturing Firms that Received Public Support

for Technology Extension Programs (percent)]

Although fiscal incentives for innovation and TEPs are conceptually different

instruments, their performance can be improved by targeting the synergies that might emerge by

combining features of both types of instruments. For example, although innovation support

programs correct market failures by providing a financial reward to the pioneers, the diffusion of
the new discoveries could be further enhanced by linking part of the reward that the pioneers

receive to dissemination activities that give followers access to technology (for example, through

training workshops, demonstration visits, or pilot productions). Furthermore, in the case of

product discoveries, pioneers’ rewards could be at least partially linked to followers’ sales of the

same product. 16

In terms of the volume of resources invested and number of firms covered, business

innovation support in Latin America is still in its infancy, as many of the programs are more in

the nature of pilots than full-fledged interventions. However, the good news is that the best

practice designs outlined in Table 3.3 are gradually being incorporated in their deployment. Of

course, the situation is far from ideal, but the trend at least in terms of instrument design seems to

be in the right direction. However, before deciding whether to expand the fiscal resources and

the coverage of these programs, the effectiveness of the existing ones in solving the

aforementioned market failures should be analyzed.

Learning from Impact Evaluations

One of the first issues to be defined in impact evaluation is the set of outcomes of

interest. A distinction should be made between input and output indicators. Input indicators are

more directly affected by the intervention: for example, total investment in innovation. To the

extent that innovation policy reduces the capital cost of firms, it could be possible to identify

which innovation policies generate an increase in innovation investment at the firm level (input

additionality). Therefore, a first-order evaluation question should assess whether innovation

policies increase firms’ investment in innovation, and the overall contribution of the private
sector to this effort: the so-called crowding-in effect. However, just assessing whether innovation

efforts increase as a consequence of a subsidy is not enough for evaluation purposes. It is also

important to assess the outputs of innovation investments by looking at variables that

demonstrate their utility such as productivity.

The traditional approach in impact evaluations of business innovation policies considers

the impact of these programs on direct beneficiaries. However, direct beneficiaries are just one

component of the social returns, and perhaps the least interesting one. A key rationale for these

policies is based on externalities. Thus, a first-order question of impact evaluation of these

programs should be the extent to which they generate spillovers. This can be done by tracing the

impact of the programs on indirect beneficiaries (such as through labor mobility or geographical

colocation). Although very few impact evaluations to date have explored the relevance of

spillovers, some of them have been done specially for this report. 17

Impacts on Innovation Investment

As in other regions, evaluating the effects on investment has been the most common

approach to impact evaluation for Latin America and the Caribbean. 18 Table 3.4 summarizes the

results of 16 impact evaluations done in the region. The evidence across the different studies is

that fiscal incentives clearly stimulate innovation or R&D investments. In almost all cases, the

evaluations found a positive and significant effect on program beneficiaries. Furthermore, in

seven evaluations, where the main impact indicator is private investment in innovation or R&D,

the results for this variable are also positive and significant, suggesting crowding-in. Fiscal

support may be having a signaling effect on the quality of the projects, allowing firms to

leverage additional resources from financial markets (see Benavente, Crespi, and Maffioli,
2007). Comparing across the different instruments, matching grants schemes were not found to

have a significantly different multiplier effect on investment than loans or tax incentives.

However, matching grants schemes clearly dominate when they provide funding conditioned on

collaboration (see cases with “Firms & UNIV” as beneficiaries in Table 3.4 cases) or when they

target new innovators. 19 Apparently, matching grants programs are particularly well-suited for

building linkages among the different actors of the innovation system, addressing both market

and coordination failures, and supporting innovation based entrepreneurship.

The majority of the studies summarized in Table 3.4 use techniques that construct

comparable groups of beneficiaries and nonbeneficiaries on the basis of observable

characteristics of the firms. 20 This provides for an accurate assessment of the programs’ selection

process, which per se provides valuable information on the programs’ targeting. The results

show that firms with high levels of human capital or previous experience in managing innovation

programs are more likely to be selected. This is not surprising, considering the weight normally

given to quality when selecting proposals. However, a system based on past accomplishments

might overlook new innovators, which may be more prone to suffer market failures. Ceilings of

maximum support per firm could be considered, given the trade-off between fostering

excellence, which may require multiple interventions for certain beneficiaries, and variety. The

matching grant instrument is particularly well-suited to balancing excellence and diversity. It is

also important to improve the coordination between innovation support and technology extension

programs, given the focus in TEPs on building innovation management capacities in firms.

[Insert Table 3.4 Effects on Innovation Investment (Input Additionally):

Testing Crowding in/ Crowding out Effects]
Impacts on Firm Performance

At the international level, fewer studies analyze the effect of public support on firm

performance; the results are mixed. The main difficulty in this type of study is that a longer time

horizon is required to detect these effects. While investment effects can be detected almost in

conjunction with the receipt of public financing, other effects are detectable only after the

innovation has taken place. Thus, rigorous impact evaluations of these effects may require

following firms for a minimum of five years after receipt of public financing. In order to fill this

knowledge gap, the Inter-American Development Bank reassessed some of the programs in

Table 3.4 over a longer period and looked at impacts on productivity. Table 3.5 summarizes the

results for the five evaluated programs. All the programs were evaluated using the same

approach, with the main output indicator being labor productivity. The results suggest significant

increases in labor productivity: from 9–12 percent when only individual firms are targeted, and

from 10–24 percent when joint firm-university projects are supported. Additional evidence

shows that strong complementarities could be achieved when the support of different programs is

combined in sequences of multiple treatments. Such complementarities are found by evaluating

the combined effects of Chile’s FONTEC, supporting individual firms, and FONDEF,

encouraging university-firm collaboration (Álvarez, Crespi, and Cuevas, 2012), or when

innovation support programs are combined with other PDP programs. 21 Complementarities also

exist between innovation policies and policies that normally form part of the framework

conditions, particularly competition policies (Box 3.4).

 [Insert Table 3.5 Output Additionality: Testing for Productivity

Impacts]

The IDB has also evaluated a series of TEPs. Castillo et al. (2014a) evaluate the effect of

Argentina’s Support Program for Organizational Change (PRE) on employment and wages. 22

The program cofinanced technical assistance to support process and product innovation

activities. Using a unique dataset with information for the population of firms in Argentina, the

study finds large effects on employment attributable to the program’s support, with increases of

around 20 percent. For the median firm, participation in the program generated five additional

jobs. The program support for process innovation increased real wages by 2 percent, while

support for product innovation increased real wages by 4 percent. The evaluation also provided

evidence of the program’s positive effect on both firm survival and export.

Benavente and Crespi (2003) analyze the impact of the Chilean Productive Development

Program (Programa Asociativo de Fomento, PROFO), which promotes joint projects among

groups of SMEs to improve access to markets and help them innovate, which sparked

productivity improvements of 11 percent vis-à-vis the control group. 23 In addition, the social

return rate of the program was at least 20 percent. TEPs oriented to the diffusion of agricultural

technology demonstrated qualitatively similar results (see Table 3.5). 24 Overall, the results

confirm that TEPs are effective in achieving their expected results and that different approaches

work when applied in the proper contexts. Obviously, evidence on effectiveness should also be

complemented by rigorous cost-benefit analysis.

 [Insert Box 3.4 Competition and the Impacts of Innovation Grants: The

Case of Chile]

The Search for Spillovers

Although knowledge spillovers are at the core of the theoretical justification for

innovation policy, very few impact evaluations measure these potential effects. This omission

probably reflects the difficulty of identifying the mechanisms through which spillovers occur. In

the context of impact evaluation, measuring spillovers implies identifying the impact of the

program not just on direct beneficiaries (firms that received program support) but also on indirect

beneficiaries (firms that received benefits from the program through their relation with direct

beneficiaries), as well as groups of comparable nonbeneficiary firms.

To fill this gap, the IDB has recently undertaken studies in Argentina (Castillo et al.,

2014b) and Brazil (Ingtec and USP Research Group, 2013). The two studies focus on fiscal

incentive programs, and define labor mobility as the main mechanism through which knowledge

spillovers occur. Because much of this new knowledge is “captured” by the human resources

operating in the beneficiary firm during the execution of the project, relevant spillovers may

occur when one of these workers moves to a different firm, carrying part of the knowledge

generated thanks to the program support.

To track and measure the effect of spillovers, both studies use administrative employer-

employee longitudinal datasets that track the mobility of qualified workers from direct

beneficiary firms to other firms (indirect beneficiaries). Once both direct and indirect
beneficiaries were identified, the causal effects—direct and indirect—of the programs were

estimated.

Findings are summarized in Table 3.6. Because of data limitations, effects on innovation

investments are available for the Brazilian programs only. The findings for Brazil show positive

spillover effects in terms of human resources devoted to innovation activities, with increases

ranging from 6–17 percent, depending on the programs. Interestingly, the program promoting

cooperation between firms and universities produces the largest spillover effects (17 percent),

providing evidence that spillovers could be larger when the knowledge generated is more

“generic.” In terms of spillovers on firm performance, the studies on Brazil and Argentina

confirm that the programs positively affected firm growth (measured in terms of employment),

with effects ranging from 7–20 percent, and exporting by boosting export probability in

Argentina by 4 percent and the ratio of exports/employees in Brazil between 17 and 23 percent.

Overall, these results strongly support the “lack of appropriability” rationale at the basis

of innovation policies. They also provide an extremely valuable input for the future social cost-

benefit analysis of these policies, providing ranges for the potential magnitude of the spillovers.

In summary, the evidence is that fiscal incentives for innovation are effective in stimulating

business innovation investments and productivity growth. This implies that the programs in

general are targeting firms that might be facing some sort of market or coordination failure;

when these constraints are relaxed, firms react favorably by increasing their innovation

investments. The evidence also shows that the spillovers generated by these programs can be

substantial, suggesting that they are effectively correcting for market failures. However, impact

evaluation results also suggest that effectiveness can be enhanced by a better focus on market

failures (for example, by targeting investments in intangibles rather than tangibles, hiring
researchers by firms or focusing on university-industry collaboration) or on those firms more

prone to suffer market failures (such as innovative SMEs and young innovators). Impacts could

be improved by seriously considering the interactions between innovation programs and

competition, among innovation programs and other PDPs (such as those for export promotion, as

discussed in Chapter 8), and between innovation programs and TEPs.

[Insert Table 3.6 The Search for Spillovers]

Keys to Success

Innovation policies are complex. They require the resolution of complicated market and

coordination failures; they involve multiple stakeholders, and require long gestation periods. For

these reasons, successful implementation requires significant institutional capabilities, including

the ability to engage with the private sector, coordinate across public agencies and guarantee the

continuity of policies. The experience from developed countries and successful catch-up

economies suggests that successful implementation of innovation policies requires building four

types of complementary institutional capacities (World Bank, 2008).

Strategy setting capabilities: In successful cases, the fundamental pillars of innovation

policy are normally agreed to at a high level within government by a multi-stakeholder

organization that sets a long-term national innovation strategy with clear and measurable

objectives and binding recommendations about the policy mix, financing, instruments,

jurisdictions, and mandates. Ideally, the institution that assumes this role should not be tied to the

political cycle, in order to assure the continuity of innovation policies, and be well-staffed with
technical resources to make strategic studies and foresight exercises. Several developed countries

have established public-private councils that fulfill this function, including Canada, Finland,

Germany, Ireland, and South Korea. In Latin America, high-level innovation councils are in

place in Argentina, Brazil, Chile, Costa Rica, Mexico, and Uruguay. However, in many cases,

these organizations are not fully institutionalized, multi-stakeholder representation is weak, they

lack operational resources and their power depends on the interest of the current administration

in power; this violates some of the key principles of these organizations. In many cases, these

institutions only play an advisory role, while line ministries still handle the strategic decisions.

Policy coordination capabilities: Effective innovation policy requires coordination

among the different ministries involved in policy implementation. In particular, policy

coordination capacities translate the long-term innovation strategy into specific policy designs

and allocate budgets to implementing agencies. Given the multiple ministries involved and

idiosyncratic organization of the public sector, two different models of policy coordination can

be identified in the region. Some countries have established “innovation cabinets,” where

different ministries discuss policy coordination and implementation (Chile and Uruguay). In

other cases, this policy coordination function is concentrated in a specific ministry that leads

cross-ministerial cabinets (Argentina and Brazil). Both models have pros and cons. An

innovation cabinet is likely to be more inclusive and have better horizontal coordination, but may

also suffer from higher transaction costs. A dominant coordinating player could be more efficient

at coordinating related agencies, but lack horizontal reach. Despite these advances, the countries

in the region still have a long way to go with regard to policy coordination, in particular

regarding the coordination of innovation policies and other PDP policies (Crespi and Tacsir,
2012) and also regarding policies implemented at different levels of government (such as federal,

regional, and municipal).

Policy implementation capabilities: The best practice from developed countries suggests

that policy implementation should be carried out by a merit-based technically capable civil

service able to partner with the private sector (Devlin and Moguillansky, 2012). Given the

current limitations of state organization in many countries, accumulating capabilities in this

dimension is not an easy task. Evidence from successful cases suggests that this is sometimes

better attained by autonomous agencies, which tend to enjoy higher stability, are more open to

experimentation, and more responsive to client needs than centralized ministerial departments

(OECD, 2011 and Box 3.1). Autonomous agencies also could have more flexibility to manage

talent and hire the human capital required to run their operations. Consistent with this trend,

several Latin American and Caribbean countries, such as Argentina, Brazil, Chile, and Uruguay,

have created implementing agencies with some of these characteristics, although the exact

agency traits vary by country. Whatever the actual institutional configuration of the agency, it

should be accountable for the efficiency of its operations at the policy and strategic levels and

should be clearly aligned to policy-level decisions. Policy implementation is more than just

managing private sector incentives for business innovation; it also encompasses policy

capabilities to align the incentives of public research centers and technological institutes to

respond to private sector needs and to simulate more pertinent technological research. 25

Monitoring and evaluation capabilities: Accountability is critical for innovation policies

to function well. More specifically, monitoring and evaluation capacities are central not only in

order to abort wrong projects early, but also to avoid private sector capture of policy

implementation. For effective monitoring, every budgetary line should be linked to measurable
outputs so that both outputs and resources can be tracked over time, and deviations from goals

can be identified and discussed. Evidence form developed countries suggests that the whole

policy system should have an evaluation arm in charge of implementing impact evaluations of

the different programs. The evaluation unit should be external (or nested in a different

organization within government) so as to assure transparency and credibility. Monitoring and

evaluation capacities are severely underdeveloped in the region to date. Although ministerial and

agency-level bodies are gradually building the capacity to conduct impact evaluations, they still

have a long way to go in order to develop program evaluation plans that incorporate state-of-the

art methodologies and, at the same time, set up the information technology systems to produce

the data needed to conduct impact evaluations.

In many cases, lack of institutional capacity translates into serious implementation

bottlenecks. First, specialized human capital to manage innovation policy is scarce. Training is

sometimes done on an ad hoc basis. In some countries, low public sector salaries induce high

rotation of personnel, which hinders efficiency and favors capture. Second, public management

and financial systems are not adequately set up to manage programs that require regular financial

transfers from the public to the private sector and flexible, timely disbursements according to

private sector needs. Third, obsolete information systems hinder monitoring and evaluation.

Fourth, there is a shortage of external examiners who do not have conflicts of interest, especially

in contexts where the research community is small. Finally, low differentiation of functions

across multiple governmental organizations results in serious problems of coordination, overlap,

and conflicts of interest among the policy actors.

 Without giving serious consideration to the institutional arrangements needed for

successful innovation policy implementation, government failures could be worse than market

failures. Countries in the region have only recently started to build up institutional capacities in

some of the above-mentioned dimensions. Not surprisingly, those countries where impact

evaluations show the most successful results are also those where at least some of the

institutional concerns have been tackled. However, this process of institutional capacity building

is still incipient for most of the countries.

One Step Forward, Two Steps Back

Although over the last 20 years, the region has made remarkable progress in improving

the framework conditions, this has not been enough to trigger a process of productivity catch-up.

Indeed, the market and coordination failures that hinder innovation investment and technology

adoption are still pervasive, and governments have notably failed to find a proper policy solution

for many of them. Nowadays, the private sector innovation investment gap between Latin

American and Caribbean and developed or successful emerging economies is larger than it was

20 or more years ago—despite the persistently high social returns to innovation investments. A

combination of investment gaps with increasing returns can only be interpreted as a sign of a

severe innovation shortfall in the region.

Despite this rather grim scenario, some glimmers of optimism are beginning to appear

throughout the region. First, after many years of inaction, several countries have started to invest

significantly in education, and in particular in disciplines related to science, technology, and

engineering. Some countries have complemented these efforts with additional investments in
research capacity and technological infrastructure. Although these efforts are relatively recent

and have been insufficient to induce an improvement in business innovation investment, they are

a step in the right direction that needs to be sustained over time.

Second, after many swings in innovation policy frameworks, an emerging consensus

finds that the new emphasis on building research capacity and human capital needs to be

complemented by a consistent stimulus on the demand side (the firms) and on solving the

coordination failures that hinder the interaction between supply and demand. In other words,

neither a supply push nor a demand pull by themselves are enough; the most likely dividends

will come from working on both sides and their interaction. The focus should be on putting in

place those policy interventions that can jumpstart the region’s innovation systems by solving

market and coordination failures.

However, in contrast with growing public support to the research base and human capital

formation, the fiscal budgets allocated to business innovation programs remain rather meager. To

some extent, business innovation policy in the region is still in its infancy. Yet as the evidence

presented in this chapter shows, when these programs are correctly designed, they are capable of

triggering additional and sustained innovation investment, increasing productivity and

employment, and generating spillovers. Although the risks of moral hazard and capture are

always present, these problems can be minimized when several basic design principles are put

into place (such as program support with a sunset clause, private sector cofinancing, allocation

through competitive processes, ex ante evaluation through external peer-reviewers, and ex post

impact evaluation.)
 Yet boosting the budgets of these programs will not be enough. In addition to basic

design principles, these programs must be specifically designed to maximize their social returns.

Issues such as increasing differentiation of the programs according to the basic innovation

capabilities of the beneficiaries need to be integrated within the policy mix. Special consideration

should be given to programs with clear incentives for collaboration among firms and universities,

since programs with such features yield a higher multiplier effect in terms of investment and

spillovers. Incentives should aim clearly toward those projects that generate spillovers and

technology diffusion.

In addition, many issues with regard to the policy mix are not being tackled properly and

need to be given serious consideration, including more experimentation. Among the key issues

that require further investigation are the effectiveness of the available instruments to diversify

the production structure by encouraging product innovation and enhancing the entry of new

innovative start-ups, the role of policy to stimulate innovation in the service sector, and the

importance of better aligning incentives with performance (not only by funding investment but

also by linking the generosity of fiscal incentives to the outcome of those investments). Finally,

and more important, far more consideration should be given to institutional arrangements that

bring about continuity, good governance, and productive public-private collaboration for

innovation. Arguably, reform of these institutional aspects lags the furthest behind in the policy

agenda. In particular, important institutional capacities, such as policy coordination, are still

weak, and others, such as strategy setting and impact evaluation, are mostly missing. This

dimension of institutional capacity building offers an important opportunity for regional

cooperation.
Box 3.1 Fostering Innovation through Government-sponsored Mission-

oriented Research and Public Procurement: The Case of the United States

Mission-oriented R&D is research funded by public agencies to support their activities. A

key feature of mission-oriented research is that policymakers, rather than scientists, choose the

fields in which large investments of public R&D funds are made. Allocation decisions are based

on assessments of the research needs of specific agency missions in fields ranging from national

defense to agriculture, health, energy, and other activities. The R&D investment budgets of most

OECD nations are dominated by programs that serve specific government missions. According

to a report by the U.S. National Science Foundation, mission-oriented research ranges from 50

percent of total government R&D spending in Germany to 90 percent in the United States, with

the budgets of Japan, France, the United Kingdom, Canada, and South Korea falling in between

those figures (National Science Board, 2006).

A key feature of mission-oriented research is that projects being funded are normally of a

more applied nature. This has led to the argument that to the extent that research findings are

very specific, the scope for spillovers is narrower than in the case of other research. Empirical

evidence has shown, however, that this is not the case, as many of the technological

breakthroughs currently used by the private sector originated in mission- oriented research

programs (including such notable examples as semiconductors, the Internet, GPS, hybrid corn,

MRIs, and hydraulic fracturing). Moreover, private sector spillovers can be maximized when

mission-oriented research tilts toward funding new bodies of scientific or engineering knowledge
that support innovation across different sectors; when it focuses on developments in the early

phases of a new technology; when it funds new publicly available technological infrastructure

(such as research labs in universities or research centers); and when the procurement rules foster

both competition and collaboration among research teams, universities, publics labs; and firms.

Mission-oriented R&D is normally complemented by substantial purchases of new

technology from one or more public agencies. Placing large orders of early versions of a

technological device allows the producer to learn, improve quality, and reduce prices for other

private sector users. Procurement rules can also promote technology diffusion. For example, in

the U.S. defense sector, public procurement is sometimes accompanied by policies that require

the supplier to develop a “second source” for the product: that is, a different domestic producer

that could manufacture a functionally identical product in order to avoid supply interruptions.

Mission-oriented research and public procurement, however, are not without risks. Large

publicly funded research programs might increase the costs of R&D (such as the salaries of

researchers), crowding out private sector R&D investment. Mission-oriented research could also

bias the research agenda toward applications that are not easily transferred to the private sector.

The United States has solved this institutional challenge by managing mission-oriented research

programs through specialized agencies. An example is the Defense Advanced Research Projects

Agency (DARPA), set up in 1957 within the Department of Defense to invest in high-risk, high-

payoff research. DARPA is a small, flexible, and flat organization with about 140 technical

professionals. It is exempted from the normal federal regulations for civil personnel, providing it

with important flexibility for managing talent. DARPA’s technical staff are hired or assigned for

four to six years. All key staff (office directors and program managers) rotate to ensure a

constant infusion of fresh thinking and perspectives. This gives DARPA the flexibility to get into
and out of an area without the burden of sustaining staff. DARPA neither owns nor operates any

laboratories or facilities. The overwhelming majority of the research it sponsors is done in

industry and universities. Project-based assignments are organized around competitions to solve

a specific technology challenge. Then the development and production are handed off to the

military services or the commercial sector.

Sources: Tether (2008); Mowery (2010); Singer (2014).

Box 3.2 Innovation Policy Building through Catch-up: The Case of South

Korea

South Korea has been one of the most successful latecomer economies in achieving rapid

economic growth and is approaching the ranks of advanced economies in terms of GDP per

capita. One element of Korea’s success has been its emphasis on capability and technological

development, which has led to the consolidation of private exporting and R&D capacity. During

the catch-up process, Korea went through four phases. In each phase, the government

implemented innovation policies using a broad range of instruments.

Initial efforts (the 1960s to the mid-1970s)

In the 1960s, when Korea began its modernization process, the country faced two key

barriers: low technological capabilities of domestic firms and poor human capital (in particular,

in applied sciences and engineering). The government focused on encouraging technology

imports with licensing, developing a new graduate school of engineering and applied sciences

(the Korean Institute of Science and Technology, KIST), and setting up key institutions for

science and technology infrastructure. These actions facilitated the absorption of imported
technologies and attracted technology-based FDI. In this stage, domestic firms participated only

in assembling and packaging processes, with very limited investment in innovation. For the

Korean firms, this was a learning-by-doing period without an explicit attempt to develop new

capabilities or technologies. During this period R&D investment was never higher than 0.5% of

the GDP.

More active catch-up phase (the mid-1970s to the mid-1980s)

In this second phase, Korean firms more actively adopted foreign technologies through

imitative innovation and reverse engineering. They invested more intensively in adapting foreign

technology and developing local technological capabilities, mainly through technological

licensing and knowledge transfer. The government focused on technological development by

funding private R&D through tax incentives, and by conducting R&D activities directly and

sharing the results with private firms. In the 1980s, a joint public-private R&D program was set

up to support higher-risk projects. Consequently, the R&D/GNP ratio increased from 0.42

percent in 1975 to 1.41 percent in 1985. During this stage, government investment in R&D still

exceeded private sector investment.

Rapid catch-up (the mid-1980s to the mid-1990s)

This third phase was a period of rapid catch-up led by the major Korean businesses.

Firms increased production of knowledge-intensive products and started to develop new

products. Realizing the limits of a strategy based on licensing and embodied technology transfer,

Korean firms established their own in-house R&D centers. To encourage this trend, the

government eased the accreditation process for setting up private R&D institutes, and a large

number of institutes were established. The R&D/GNP ratio increased from 1.41 percent in 1985

to 2.32 percent in 1994. Such active engagement of private R&D activities enabled Korea to
absorb the newly emerging technologies. From this period onward, private R&D investment has

been a key part of the Korean innovation and technology development process, accounting for

more than 70 percent of total R&D investment.

Maturing of the catch-up phase (the mid-1990s to the present)

As South Korea approaches the technological frontier, the country is entering a new and

critical phase in its development. With the slower growth of labor and capital inputs and

increasing competition from new industrializing countries, South Korea faces new challenges.

The catch-up model is now under stress, and South Korea is shifting from a “catch-up” to a

“creative” innovation system. The creative model requires increased spending on R&D – by both

public and private sectors – and improved knowledge flows and technology transfer across the

system. This demands stronger support to innovative SMEs and Start-Ups; increasing the role of

longer-term, fundamental research; developing research capacity in the universities; and dealing

with lagging productivity in services. This transition towards a creative economy can already be

seen in some innovation indicators. Patents owned by Koreans increased from 7 in 1982 to 3,558

in 1999 according to a U.S. register. In 2006, the R&D/GNP ratio passed the 3 percent threshold.

Sources: Lee (2013) and OECD (2009).

Box 3.3 FONTAR’s Toolkit: Basic Rules for Allocating Subsidies vs. Credit

The Argentinean Technological Fund (Fondo Tecnológico Argentino, FONTAR) was

created in 1995 and has been one of the pillars of Argentina’s innovation policy. Although the

program has expanded its interventions over time, it has focused on providing support to
business innovation projects through two main instruments: reimbursable funding, through credit

for innovation; and nonreimbursable funding, through matching grants and tax credits.

The assessment of innovation projects usually requires very specific technical expertise.

The lack of such expertise can worsen the problem of asymmetric information between investors

and the innovator. Programs such as FONTAR, which assess innovation projects through

appropriate review processes, provide valuable signals to financial markets about the technical

and commercial sustainability of the investments. Moreover, innovation projects are riskier,

more intangible, and more prone to spillovers than technology adoption projects, which are

based on more easily collateralized physical investment (like machinery). For this reason,

external investors systematically require higher risk premiums to finance innovation activities, or

simply avoid funding them.

Because FONTAR’s lines of funding target different kinds of investments with different

degrees of risk and lack of tangibility, the instruments used for each line can be sharply

differentiated. Thus, their nonreimbursable instruments specifically target R&D projects. By

contrast, projects aimed at the adoption of existing knowledge embedded in tangible assets are

less risky and the returns are more likely to be appropriated by the innovator. Thus, the policy

intervention focuses on solving a problem of asymmetry of information by providing credit. The

table that follows shows how FONTAR’s officers allocate the different projects to the different

instruments.

[Insert Table B 3.3.1 Basic Rules to Allocate Funding to Innovation Projects:

The Case of FONTAR]

Box 3.4 Competition and the Impacts of Innovation Grants: The Case of Chile

The Chilean government has been experimenting with business innovation policies since

the early 1990s, mostly through different direct support (matching grants) programs to stimulate

business innovation and university-industry collaboration. Most of these programs have been

managed by the National Development Agency (CORFO) and the National Science and

Technology Council (CONICYT). More than 6,000 projects have been funded since 1991. This

experience has generated rich evidence to learn about a variety of impacts of innovation policies

which can be exploited to maximize the impacts of policy.

One long-standing issue concerning innovation is the relationship between innovation

and competition. It has been argued that innovation is at odds with competition because the need

to generate innovation rents to reward the innovators normally implies accepting some sort of

market distortion (for example, through the granting of intellectual property rights) as the price

to pay to obtain more innovation. Recent research on this subject has reevaluated this view,

finding that the relationship between these two variables is more complex than previously

thought. Aghion et al. (2012) argue that the effects of fiscal incentives to stimulate innovation in

a given sector vary depending on the degree of competition among firms: the more competitive

the sector is, the more these firms will be encouraged to innovate in order to escape competition.

In other words, a certain demand for innovation is necessary in order for fiscal incentives to be

effective, and competition is the trigger for this demand. Using data from China over the 1988–

2007 period, Aghion et al. (2012) report results consistent with this view.
 In order to explore whether this argument applies to Chile, the Chilean data on the

beneficiaries of business innovation programs were linked to the manufacturing census over the

1991–2006 period. A difference-in-difference approach tested whether innovation programs have

had any positive effect on firms’ total factor productivity. To see whether these effects vary

across sectors depending on the intensity of competition, the treatment variable was interacted

with an index of competition calculated at four-digit industry levels.

The results of these interactions are plotted in the figure below, where sectors are ranked

from a very low level of competition on the left to a very high level of competition on the right.

As the figure shows, the impacts of business innovation programs clearly grow with the degree

of competition in the sector. In fact, the impact may have been negative in sectors with very low

levels of competition. These results have strong implications for innovation policy design: the

impacts of the program could have doubled if the fiscal support focused only on those sectors

with a high intensity of competition. This also points to strong complementarities between

innovation and competition policies.

[Insert Figure B3.4.1 Innovation Policy and Competition: Evidence

from Chile]
Figures
 Figure 3.2 Social Returns to R&D, Latin America and the Caribbean vs. the OECD

0.6
Social return rate

0.5

0.4

0.3
1960 1970 1980 1990 2000 2010

Typical OECD country Typical LAC country

Source: Authors' calculations based on Griffith, Redding, and Van Reenen (2004).

𝑝
𝑅 𝑅 𝑅 𝑅
𝑖𝑡∆𝐴 1 =ƿ − +δ 𝑙𝑛 −
1 +∑3 δ −
ƿ=1 2𝑝 x 𝑙𝑛 −
Note: Estimates are based on the following model: 𝑌 𝑖𝑡 −1 𝑌 𝑖𝑡 −1 𝑌 𝑖𝑡 −1 𝑌 𝑖𝑡 −1, where A is total factor
productivity, R is research and development investment, and AF is the productivity frontier. In this model, the social return rate to
R&D has two sources: its contribution to innovation, which that pushes 𝑅
the technological
𝑅
𝑝 frontier; and the creation of absorptive
capacities for technology transfer. In other words, 𝑆𝑅𝑅𝑖𝑡 = ƿ1 + ∑3ƿ=1 δ2𝑝 − x 𝑙𝑛 − . The results are based of OLS estimates.
𝑌 𝑖𝑡 −1 𝑌 𝑖𝑡 −1
Results using instrumental variables (with the Ginarte and Park IPR protection Index and the Productivity Frontier as instruments)
were qualitatively similar. OLS are more conservative than IV estimates.
Table 3.1 Explaining the Gaps in Business R&D: The OECD vs. Latin America and
the Caribbean (percent of the total GAP)

Time period 1980–95 1995–2010

Business R&D % of GDP (gap between
0.9 1.18
OECD and Latin America)
Knowledgea 29.7 23.7
Human capital 23.2 25.8

Financial development
26.9 15.3
Production structure
10.7 26.0
Residual 9.5 9.2
Total 100 100

Source: Authors' calculations based on Lederman and Saenz (2005); OECD (2010); World Bank (2010b); IDB (2010);
RICYT (2013).
Note: The contribution to the gaps is based on a model, 𝑦𝑖𝑡 = 𝑋𝑖𝑡 β𝑡 + 𝐿𝐴𝐶𝐼 δ𝑡 + τ + ε𝑖𝑡 where Y is business
R&D as a percentage of GDP, X is a vector of variables described in the text, and LAC is a dummy variable
identifying Latin American and Caribbean countries, capturing a residual gap that is not explained by the above-
mentioned variables.
a. The main variables are measured as follows: (i) knowledge is built by accumulating publicly performed R&D per
worker; (ii) human capital is approximated by average years of schooling in the labor force (due to data constraints,
human capital is not corrected by quality); (iii) financial development is measured by the intensity of credit to the
private sector; and (iv) production structure is measured by using the proportion of high-tech manufacturing on
value added.
Table 3.2 Innovation Policies in Developed Countries: A Taxonomy

TYPE
Horizontal Vertical
Higher education/Training. Technological institutes
Support to scientific research. (agriculture, industry, energy,

Public good
Intellectual property rights. fishing, etc). Standardization.
Research infrastructure. Human Thematic funding. Signalling
capital inmigration. Labor training. strategies. Information diffusion
S
Competition policy. Regulation. policies (extension systems).
C
Technology transfer organization. Technological consortiums.
O Contests
P
R&D subsidies, R&D tax credits. Public procurement. General
Market intervention

E Financial measures (guarantees for purpose technologies (ICTS,

technology invesments, biotech, nano-tech). Strategic
intangibles values, etc). Adoption sectors (semiconductors, nuclear
subsidies. energy, electronics, etc). Defense
sector.

Source: Authors' compilation.

Table 3.3 Fiscal Incentives: Subsidies vs. Tax Incentives

Direct subsidies Tax incentives

Mechanism Project-based financing. Based on firm-level R&D activities.

Impacts Reduce marginal cost of R&D activities. Reduce marginal cost of R&D activities.

Collaboration Funding can be targeted toward collaboration. Deduction can also target collaboration.
Fully market friendly mechanism. The firm decides. It might be biased
Spillovers Funding can be targeted toward projects with high spillovers.
towards more privately appropriable projects.
Funding can provide partial cash advances (relaxing liquidity They operate fully ex post, and are less suitable to solve financial
Liquidity constraints
constraints). constraints.
Funding can provide "signalling" to external investors. No signalling.
High (funding can be targeted toward firms with innovation problems, Low (effectiveness depends on the general tax context of the country,
Focus
such as innovative SMEs or start-ups). other tax exemptions and loopholes) and biased toward larger firms.
Implementation costs High ex ante and ex post. Low ex ante but high ex post.
Institutional capacities High capabilities ain innovation agencies. Lower capabilities in innovation agencies, but higher in tax authority
Firm capacities High (to formulate a project) High (to identify an innovation program)
Uncontrolled, the fiscal costs depend on decisions taken by the firms.
Fiscal costs Controlled costs and transparent. Funding targets the marginal project. When based on volume, subsidies go to intra marginal projects as
well.They make the fiscal system more complex.
They create an incentive to artificially classify non R&D expenses as
Moral hazard Financing can go to firms that do not face market failures.
R&D which may not be easy to control for tax authorities.
Implementation of cofunding schemes (matching grants) with nominal Base the incentives on R&D growth rather than volume or establish a
limits and a list of elegible expenses. project-based decision making process similar to subsidies.
Subsidy rate proportional to the size of spillovers (e.g. higher in public
Good design practices goods, generic research or collaborative projects). Build monitoring and evaluation capacities in the tax authority
Include in the deduction a premium for externalities (e.g. collaboration or
Implementation of a competitive allocation process (call for proposals). the hiring of R&D personnel).
Transparent allocation by a public-private council based on evaluations
Inclusion of carry-forward provisions or cash conversions for new firms.
by external and independent peer reviewers.
Capacity building in firms for project formulation and setting of Predict the fiscal cost and inlude it in the budget, setting a transparent
milestones for funding. mechanism to allocate the credits when the demand is higher than supply.

Include a sunset clause with a careful monitoring and evaluation system. Include a sunset clause with a careful monitoring and evaluation system.

Source: Authors’compilation based on Lederman and Saenz (2005); OECD (2010); World Bank (2010b); IDB (2010); RICYT (2013).
Table 3.4 Effects on Innovation Investment (Input Additionality): Testing for Crowding-in/Crowding-out Effects

Country Evaluation Program Inte rve ntion Be ne ficiarie s Indicator Impact Crowding Me thod
pe riod name in/out
Argentinaa 1994-2001 FONT AR-T MP1 Subsidized loan Firms Ln (total R&D) 0.15** In FE-IV
Argentinab 1998-2006 FONT AR-ANR Matching grants Firms Ln (private innov exp) 0.18* In FE-CS
Panamac 2000-2003 FOMOT EC Matching grants Firms Ln (T otal R&D) 0.15** No evidence FE-CS
Uruguay d 2000-2006 PDT -I Matching grants Firms Ln (private innov exp) 0.84** In FE-CS
Mexico e 2004-2007 EFIDT R&D tax credit Firms Ln (private R&D) 0.25** In FE
Colombiaf 2000-2002 T ax Incentives R&D tax credit Firms Ln (private R&D) 0.06** In SM
Argentinag 1995-2001 FONT AR CFF R&D tax credit Firms Ln (private R&D) 0.13*** In FE
Brazilh 2005-2010 LEI-DO-BEM R&D tax deduction Firms Ln (R&D employment) 0.07*** in FE
Brazilh 2001-2008 LEI da Informatica R&D tax deduction Firms Ln (R&D employment) 0.01 out FE-CS
Argentinai 1994-2004 FONT AR CFF R&D tax credit Firms (T otal R&D $) 1.90** In SM
Argentinaj 2001-2004 FONT AR-ANR Matching grants Firms (T otal R&D intensity) % 0.18** No evidence DID-PSM
Brazilk 1996-2003 ADT N Subsidized loan Firms (Priv. R&D intensity) % 0.66** In PSM
Brazill 1999-2003 FNDCT Matching grants Firms & UNIV (Priv. R&D intensity) % 1.63** In PSM
Chilem 1998-2002 FONT EC Matching grants Firms (T otal R&D intensity) % 0.74* Partial out DID-PSM
Panaman 2006-2008 SENACYT Matching grants Firms (T otal R&D intensity) % 0.13** In PSM
Colombiao 2002-2003 COFINANCIACION Matching grants Firms & UNIV (T otal R&D Intensity) % 1.20* In PSM

Source: Authors’ compilations based on the studies noted. aChudnovsky et al. (2006). b López, Reynoso, and Rossi (2010). cMaffioli, Pusterla, and Ubfal
(2011). d CENIT and CPA Ferrere (2010). eCalderón-Madrid (2011). fMercer-Blackman (2008). g Binelli and Maffioli (2007). h Kannebley and Porto (2012).
i
Giuliodori and Giuliodori (2012). j Chudnovsky et al. (2006). k de Negri, Borges Lemos, and de Negri (2006a). l de Negri, Borges Lemos, and de Negri
(2006b). mBenavente, Crespi, and Maffioli (2007). n Crespi, Solís, and Tacsir (2011). o Crespi, Maffioli, and Meléndez (2011).

Note: FE-IV (Fixed Effects, instrumental variable), FE-CS (Fixed Effects and common support), FE (Fixed Effect), SM (Structural Modelling), DID-PSM
(Difference in Difference, Propensity Score Matching), PSM (Propensity Score Matching). In the case of the evaluation of SENACYT-Panama, total R&D
intensity is computed as R&D as a fraction of total innovation sales. UNIV stands for universities. *** 1% significance level, ** 5% significance level, *
10% significance level.
Table 3.5 Output Additionality: Testing for Productivity Impacts

Country Evaluation Program Intervention Beneficiaries Indicator Impact Method

period name
Colombiaa 1995-2007 COFINANCIACION Matching grants Firms & UNIV Labor productivity 0.15*** FE-CS
Colombiab 2001-10 T AX INCENT IVES R&D deduction Firms Labor productivity 0.06*** LDV
Chilec 1998-2006 FONT EC Matching grants Firms Labor productivity 0.09*** FE
Chilec 1998-2006 FONDEF Matching grants Firms & UNIV Labor productivity 0.12*** FE
Chilec 1998-2006 FONT EC only Matching grants Firms Labor productivity 0.06 FE-CS
Chilec 1998-2006 FONDEF only Matching grants Firms & UNIV Labor productivity 0.10*** FE-CS
Chilec 1998-2006 FONDEF+FONT EC Matching grants Firms & UNIV Labor productivity 0.24*** FE-CS
Panamad 2000-03 FOMOT EC Matching grants Firms Labor productivity 0.13* FE-CS

e
0.19***
Argentina 1996-2008 PRE T EP Firms Ln(employment) FE-CS
0.22***
0.02***
Argentinae 1996-2008 PRE T EP Firms Ln(wages) FE-CS
0.04***
f
Mexico Add PNAA T FP-Full subs. Firms Ln(wages) 0.05*** FE-CS
Mexico f Add CIMO T FP Firms Ls(sales) -0.05*** FE-CS
Perug Add BONOPYME T FP Firms Ln(sales) 0.16*** FE-CS
Perug Add CIT E-Calzado T FP Firms Ln(sales) No effect FE-CS
Colombiah Add FOMIPYME T FP Firms Exports 0.40*** FE-CS
Chilei Add FAT T FP Firms Ln(wages) 0.09*** FE-CS
Chilei Add PROFO-PDP T FP Firms Ln(wages) 0.08*** FE-CS
Argentinaj 2002-06 PROSAP T EP-Full subs. Grape producers Probability of adopting new variety 0.03** FE

k
14**
Uruguay 1999-2006 PREDEG T EP-Part subs. Fruit producers Adoption of new varieties FE-CS
9.3*
l
Uruguay 1999-2006 PREDEG T EP-Part subs. Fruit producers Density of plantation 108.5** FE-CS
Uruguay m 2001-03 LPP T EP-Part subs. Livestock producers Adoption of managerial practices 25.3**/18.74** FE-CS

Note: FE-CS (Fixed Effects and common support), FE (Fixed-Effects). UNIV stands for universities. Full subs stands for full subsidies. Part sub
stands for partial subsidies. *** 1% significance level, ** 5% significance level, * 10% significance level.
Source: Authors’ compilations based on the studies noted. aCrespi, Maffioli, and Meléndez (2011). b Parra Torrado (2011). cÁlvarez, Crespi, and
Cuevas (2012). d Maffioli, Pusterla, and Ubfal (2011). eCastillo et al. (2014a). fLópez-Acevedo and Tinajero-Bravo (2011). g Jaramillo and Díaz (2011).
h
Duque and Muñoz (2011). i Tan (2011). j Maffioli et al. (2011). k Maffioli et al. (2013). l Maffioli et al. (2013). mLópez and Maffioli (2008).
Table 3.6 The Search for Spillovers

Country Evaluation Program Intervention Beneficiaries Indicator Impact Method

period name
Argentinaa 1998-2006 FONT AR Matching grants Firms ln(employment) 0.20*** FE-CS
Argentinaa 1998-2006 FONT AR Matching grants Firms Export probability 0.04** FE-CS
Brazilb 2005-10 LEI DU BEM R&D tax credit Firms ln(R&D employment) 0.06*** RE-DYN
Brazilb 2006-10 LEI DU BEM R&D tax credit Firms ln(employment) 0.08*** RE-DYN
Brazilb 2006-10 LEI DU BEM R&D tax credit Firms Export/Employees 0.16*** RE-DYN
Brazilb 2006-10 Subvention Matching grants Firms ln(R&D employment) 0.15*** RE-DYN
Brazilb 2006-10 Subvention Matching grants Firms ln(employment) 0.06*** RE-DYN
Brazilb 2006-10 Subvention Matching grants Firms Export/Employees 0.23** RE-DYN
Brazilb 2000-10 ADT EN/FINEP Grants/Credit Firms-Univ. ln(R&D employment) 0.17*** RE-DYN
Brazilb 2000-10 ADT EN/FINEP Grants/Credit Firms-Univ. ln(employment) 0.07*** RE-DYN
Brazilb 2000-10 ADT EN/FINEP Grants/Credit Firms-Univ. Export/Employees 0.17* RE-DYN

Notes: FE-CS (Fixed Effects and common support), RE-DYN (Random Effects with dynamics). *** 1% significance level, ** 5% significance level, *
10% significance level.
Source: Authors’ compilations based on studies noted. aCastillo et al. (2014b). b Ingtec and USP Research Group (2013).

Table B3.3.1 Basic Rules to Allocate Funding to Innovation Projects: The Case of FONTAR

R&D Technological modernization Investment

Risk
Characteristics

Tangibles

Intangibles

Product development Purchase of high-tech equipment Industrial plant

Activities

Product innovation Automatization

Purchase of equipment for
Process innovation the new plant
Management and quality control
technologies
Applied research Production scaling up
Pilot plant Vertical integration
Instrument

Not eligible for FONTAR

Matching grants Credit
financing

Source: FONTAR (Authors' translation): http://www.agencia.mincyt.gob.ar/upload/Presentacion_FONTAR.pdf

 Figure B 3.4.1 Innovation Policy and Competition: Evidence from Chile
0.07
Average treatment effect on treated - ATT

0.05

ATT across the sample

0.03

0.01

-0.01
1 2 3 4
Quartiles of the competition index based on Lerner index: 1-low to 4-high

Source: Authors’ calculations based on Chile’s National Manufacturing Census

(ENIA, 1991–2006) and CORFO/CONICYT beneficiaries’ registers.
Note: Competition is measured by 1 minus the Lerner index, which in turn is
calculated as the ratio of operating profits minus capital costs over sales.
Endnotes

1
This is an average figure. Not all countries report these figures for every year. So for some countries, data begin in
1999. In others, data end in 2008.
2
Once produced, new knowledge can be used simultaneously by many different firms because the new “blueprints”
are not normally associated with physical constraints. This characteristic is an extreme form of decreasing marginal
costs as the scale of use increases: although the costs of the first use of new knowledge may be large, in that it
includes the costs of its generation, further use can be done at negligible small incremental costs (Aghion, David,
and Foray, 2009).
3
The nonexcludable nature of knowledge refers to the difficulty and cost of trying to retain exclusive possession of
it while, at the same time, putting it to use.
4
Technological knowledge is also more likely to be protected by intellectual property rights (IPRs). IPRs provide
innovating firms the right to temporarily exclude others from using a new idea commercially so the originators can
appropriate the rents of their investments in innovation. In exchange for this, the owner must disclose the invention
so anyone can improve upon it. However, IPRs can also generate unintended consequences, as they cause a static
market distortion in the form of monopoly power and slower technology diffusion for producers that must pay a
higher cost to transfer protected technology. In other words, IPRs also create market distortions that might or might
not be compensated by the increased incentives to innovate (De Ferranti et al., 2003).
5
It has also been suggested that traditional R&D statistics do not capture the innovation effort carried out by natural
resource–intensive industries well. For example, investments in mining prospecting are not considered part of R&D,
so mining is an activity that is more knowledge-intensive than the official statistics suggest. Given that the bulk of
Latin American countries are very intensive in these activities, this measurement problem may overestimate the gap
between the typical Latin America country and say, South Korea. However, when the comparison is done with
developed countries well-endowed with natural resources, such Australia and Canada, the investment gaps remain
and are still significant. Accordingly, the natural resource curse is not the explanation for the gap.
6
In both periods there is a residual not explained by those variables included in the analysis. Several omitted factors
are included in this residual. For example, Katz (2001) puts forward the hypothesis that macroeconomic volatility
typical of the Latin American and Caribbean development process might have affected entrepreneurs’ “animal
spirits,” making them reluctant to spend on highly risky and sunk investments such as R&D. An alternative
hypothesis could be the differences in the size distribution of firms. Unfortunately, the authors lack detailed
comparable information of R&D by size of the firm, and thus could not include this factor in the analysis.
7
An early example of this is the U.S. decision to upgrade agricultural productivity through the land-grant
universities during the second half of the 19th century (with the passing of the Morrill Act of 1862 by Congress).
Another example is the establishment of Israel’s Institute of Technology (Technion) during the early 1920s.
8
Such as energy technologies, microelectronics, aerospace, health and, in some cases, the defense sector.
9
Such as the National Institute of Industrial Technology (INTI) and the National Institute for Agricultural
Technology (INTA) of Argentina, the Agricultural Research Corporation (EMBRAPA) of Brazil, the Technological
Institute (INTEC) of Chile, and the Institute for Industrial Technology and Technical Norms (ITINEC) of Peru.
10
The fact that knowledge is a public good does not necessarily mean that it needs to be provided by the public
sector, at least in its entirety.
11
Conditional loans are a risk-sharing financial instrument by which loans could be partially or even totally written
off on the basis of three criteria: the success of the investment, the technological risk, and spillovers of an innovation
project. In Israel, for example, the repayment of conditional loans is done through royalties of between 2 percent to
5 percent of the sales of the innovated product until the original grant is fully repaid.
12
A nice feature of subsidies is that they can also provide signaling on the quality of an externally assessed
innovation project. Tax incentives are normally ex post, so they are less suitable to alleviate financial constraints and
have no signaling power.
13
Several evaluations in developed countries suggested that indeed this could be the case (Irwin and Klenow, 1996;
Branstetter and Sakakibara, 1998; Czarnitzki and Fier, 2003).
14
In the OECD (2005b) framework, the concept of “fully developed outside” is close in spirit to the idea of
“innovation new to the firm.”
15
See Ezell and Atkinson (2011) for a review on this topic. The earliest examples of technology extension programs
are found in agriculture both in continental Europe and the United States (Steinmueller, 2010).
16
In a similar line, several developed countries have put in place fiscal incentives (mostly in the form of a tax
deduction) on those profits generated from IPRs royalties or sales (see, for example, the UK Patent Box act)
17
See, for example, Lach, Parizat, and Wasserteil (2008) and Mohnen and Lokshin (2010).
18
For evidence on the effectiveness of business innovation programs in developed countries, see the summaries by
David, Hall, and Toole (2000) and Westmore (2013).
19
Chudnovsky et al. (2006) found that crowding-in effects were particularly strong in the case of new innovators.
20
In particular, propensity score matching (PSM) techniques.
21
Chapter 8 discusses the potential benefit of combining and sequencing innovation and export promotion
programs.
22
This evaluation links beneficiary data with Argentina’s social protection dataset. This allows the beneficiaries and
controls to be tracked over a long period of time at low cost. The trade-off is that productivity as such cannot be
measured. In this chapter, productivity is approximated by the average wages paid by the firm.
23
For more on PROFO, see Chapter 7.
24
This is particularly true when measures of technology adoption such as the one reported in Table 3.5 are
considered. Less conclusive are the findings on yields. In many cases, yields seem not to be affected in the short run
(which, in some cases in the only time frame considered in the evaluations), but increased over longer periods. The
available evidence seems to confirm that also for agricultural TEPs, positive effects on productivity require a certain
period of gestation; in the short run, producers may experience some adjustment costs to the new technology and
practice, which may lead to null or even negative short-run effects on productivity.
25
Given that spillovers surround scientific and generic technological research, there is clearly a role for public
research and technology centers. However, without the right institutional arrangements, these organizations may end
up being captured by scientific elites and carry out their activities in complete isolation from societal needs
(Artopoulos and Navarro, forthcoming). There is some consensus that successful public research and technology
centers should have a funding mechanism that allows them to build capabilities over the long term but at the same
time also connect their research agenda to the needs and demand of the private sector. This can be achieved by
combining long-term core funding regulated by performance agreements negotiated every four to five years with
competitive finance though matching grants funds and authorizations to provide contract research to the private
sector. Researches should also be promoted based on performance and enjoy at least some of the benefits of the
intellectual property that they are generating. Finally, and most importantly, the private sector should be represented
on the boards that control these institutions (Maloney and Perry, 2005).