Many different factors need to come together for an innovation to be successful, and as technologies and economies evolve, different policy strategies and interventions are often required. As discussed above, innovation studies emphasize that policy needs to consider the entire system that influences technological development. Monitoring an innovation system involves understanding the roles of different innovation players such as universities, firms, associations, and users; considering the entire lifecycle of a technology; analyzing multiple innovation activities such as mobilizing financial and human resources, developing sectoral and regional innovation strategies, and building new networks to exchange knowledge; as well as examining how existing economic structures block or enable new technological developments (Bergek et al. 2008; Hekkert et al. 2011). Gaps such as missing players (e.g., lead users), weak innovation activities (e.g., missing interactions), or a lack of support during a critical moment in a technology's evolution can contribute to innovation failures. A policy mandate that is too restrictive could rule out promising technological options or policy strategies.

It is also important to tailor policies to particular technological circumstances. For instance, Bonvillian and Weiss (2009) note that market-entry challenges differ in sustainable energy compared to other frequently targeted sectors such as information and biological technologies. These latter technologies can enter entirely new economic spaces or frontiers with high-price products and then gradually reduce prices as technologies mature. In contrast, without policy actions, energy technologies have to compete on price immediately because the end product (e.g., heat, light) has the same characteristics regardless of the upstream technology used. Energy technologies also enter a “legacy” sector with structural lock-in characteristics (e.g., infrastructures and regulatory systems) designed to complement the old technologies and exclude new ones. Thus the market-entry stage of development must not be neglected by innovation policy targeted to sustainable energy.

A comprehensive policy approach would consider how the entire system can help overcome market-entry barriers. Policies such as feed-in tariffs could help sustainable technologies gain market access. Furthermore, market-entry needs could help inform challenges for basic science,² supply chains could be examined to help reduce prices, and policy-makers could grapple with infrastructure integration challenges. A good example is the US “SunShot Initiative,” which undertakes a variety of activities focused on the market-entry goal of making solar energy cost-competitive with traditional energy sources before 2020 (US Department of Energy 2016).

The public-administration challenge presented by clean-innovation policy is that multiple policy tools might be relevant depending on unique sectoral, technological, and regional contexts, and policy-makers will need to change policy strategies over time as technological systems evolve. It is not clear which public-administration configurations best provide effective and comprehensive policy support. It is unlikely, and potentially undesirable, for one government department to have all the powers required to implement a tailored and comprehensive policy. Different sections of government might be best placed to implement innovation policy at any one time. There are historic examples of powerful pilot agencies leading industrial development missions, such as in Japan and Korea (Johnson 1982; Chang 1994). In the United States and Ireland, authors describe a more “networked” state with a decentralized structure of labs, agencies, departments, and specialized policy initiatives at multiple levels of government (Block 2008; O'Riain 2004). In this configuration some redundancy between government functions can be beneficial since it increases the chances of new ideas finding a space within the public sector. Further insights into achieving both coordination and policy innovation come from Breznitz and Ornston's (2013) case studies in Israel and Finland. They found that agencies with limited power and resources, yet a high degree of freedom and flexibility, produced new policy models that were picked up by other sections of government at critical moments. Their findings emphasize the benefits of creating entities within government with independence (see below) and with an ability to move across administrative boundaries.

While policy implementation might require the use of organizations with specific skills and competencies, discovering the right policy interventions requires a wide analytical purview and an ability to adapt when confronted with particular innovation problems. To promote a comprehensive innovation systems perspective, Sweden created an agency called Vinnova. Foresight exercises and innovation systems analysis play a strong role in informing the agency's decisions on where to direct policy efforts and how to advise government (see Chaminade and Edquist 2006; Jacob 2006).

Achieving comprehensiveness will likely require the participation of multiple entities within government. The question of “who does what” might depend on a given jurisdictional context and pre-existing organizational and bureaucratic competencies. Innovation theory and case examples suggest that each organization engaged in promoting innovation should have a comprehensive understanding of the innovation system and the role they can play within it, as well as an ability to solve innovation policy problems using a variety of policy tools.

Innovation is an ever-changing process beset with significant uncertainty. Policy should be able to adapt to new knowledge and shifts in industrial dynamics by changing policy designs and methods of implementation. Technological and societal changes can also create time-specific windows of opportunity that policy-makers and entrepreneurs need to identify and exploit (Sartorius and Zundel 2005). Public agencies should be able to scale up successful projects. It is also crucially important that the public sector have the capacity to cut off non-performing projects and change strategies in light of new information.

A relevant example of the importance of a flexible and adaptive approach comes from Japan's successful development of hybrid electric vehicles (Åhman 2006). Japan's Ministry of International Trade and Industry (MITI) defined an ambitious agenda to develop battery electric vehicles in the 1970s. Few of the technology targets were met; however, the electric drive-train technologies developed in search of developing a battery electric vehicle were used by car companies to produce hybrid vehicles. When government departments became aware of the hybrid technology, they quickly extended purchase subsidies toward these vehicles, creating an initial market. Thus, R&D programs created technological learning in unexpected ways, and the government's flexibility in face of these unexpected technological developments helped Japan to successfully develop hybrid vehicles.

Innovation is a long-term process that is characterized by significant uncertainty. An agency working on the frontiers of low-carbon innovation needs to be accepting of technological failures, recognizing them as normal occurrences that can contribute to learning (see Rodrik 2014). If low-carbon innovation policy is caught in what public administration scholar Donald Savoie (2015) calls “blame games,” decision-making will lose focus on fostering long-term transformations. Undue political influence might also prevent the timely cut-off of failing projects. Lipsey and Carlaw (1998) find examples where political optics and the search for prestige led to continuation of support for projects that should have been cancelled. The Solyndra case in the United States demonstrates that expending political capital on one project can create liabilities for politicians, cause administrators to continue supporting a project when it should be cut off, and direct public attention away from the overall success of broader policy initiatives.³

Autonomy from short-term political pressures is a common attribute in many innovation policy success stories, yet it is achieved in different ways. In Japan, sector development policy was housed within the powerful Ministry of International Trade and Industry (MITI). This ministry had a high degree of political autonomy because Japan had a culture of bureaucratic dominance over politics, where civil servants were held in very high prestige (Johnson 1982). In Israel and Finland, Breznitz and Ornston (2013) explain how innovation agencies achieved political autonomy because they were “low-profile” and were on the “periphery” of the public sector. The benefit of this peripheral position was that these agencies had flexibility and room to experiment with new policy approaches, and they avoided capture by powerful interests.

The drawback of an arm's-length or peripheral position within the public sector is that an agency does not have the ability to wield a full suite of policy tools.⁴ It is also possible that outsourcing certain policy functions can reduce the competence and capabilities of other government departments (Mazzucato 2015). Breznitz and Ornston's (2013) case histories explain that peripheral agencies played the special function of introducing new policy models, which were scaled up by other departments of government. Thus other areas of government, and the ability of peripheral agencies to coordinate with them, were important.

The contrast between Japan, Israel, and Finland highlights that a country's political culture determines how political autonomy is best achieved. In Canada, Savoie (2015) describes a condition in which political leaders and civil servants can be very intolerant of risk, which creates a short-term rather than long-term focus. Keeping certain innovation policy functions at arm's length could create a space for policy experimentation while also shielding political leaders from what Savoie calls “blame games.”

Authors highlight the importance of having clear objectives for policy implementation as well as a mobilizing mission (see Narayanamurti, Anadon, and Sagar 2009). This mission should be grounded in a clear understanding of the unique role the public sector plays within innovation systems. Mazzucato (2015) emphasizes that in contrast to the private sector, the state's role within innovation systems is to tackle large societal challenge and to “think big.” She argues that government innovation activities can help create entirely new economic paradigms, such as a green paradigm shift.

The clarity of the mission and policy objectives is another theme discussed in the literature. Lipsey and Carlaw's (1996) review of numerous case studies of successful and unsuccessful innovations highlights the fact that multiple objectives are dangerous because introducing a variety of non-technological objectives can be used to maintain projects that should be cancelled. Likewise, Rodrik (2014) notes that political leaders often justify policies based on numerous rationales, such as pollution reduction, jobs, and competitiveness, but these multiple goals cloud policy direction and make it difficult to know when a change of course is warranted. Politicians find it useful to justify the introduction of a policy based on multiple objectives, but these objectives could come back to haunt them because they encourage Savoie's “blame games” during policy implementation.

A benefit of clean innovation is that it is clearly associated with a public-purpose mission of promoting environmental improvements. To meet the goals related to sustainable development and a low-carbon economy, a large-scale transformation of society, or major shift in economic paradigm, is likely required (Freeman 1992). This transformative goal is distinct from more general innovation policies that are principally focused on economic growth or productivity. Transformative innovations can create disruptions in other economic sectors and induce larger structural changes (Christensen 2013; Freeman and Perez 1988). Innovation policies geared toward the general economy might have a bias toward supporting incremental innovations in existing economic sectors, while clean innovation aims to shift economic trajectories and create new sectors. Thus “innovation” and clean innovation can be quite distinct (Alkemade, Hekkert, and Negro 2011). A clean innovation mission should be recognized as separate and unique from more general innovation policy objectives. If the paradigm is truly shifting, clean innovation will become a goal that aligns all other areas of policy.

The second section of this paper discussed a lack of information as a reason that governments might fail to implement successful innovation policy. This information problem can be alleviated by ensuring consistent relationships and information exchange with organizations outside of government. In Peter Evans's (1995) seminal study on the computer industry, he discusses the need for the government to be “embedded” within the private sector. Consistent, sustained linkages with the private sector increase the government's understanding of the sectors it aims to influence. Policy-makers need to understand business strategies and anticipate how private companies will respond to a policy change. Embeddedness allows for a continuous negotiation of goals and the development of common projects, enabled by the buildup of trust, mutual understanding, and reciprocity with respect to information exchange. An embedded public sector is not only smarter but also increases its options for policy implementation because soft instruments such as strategic planning, coordination, regulatory changes, and even suggestion can become more effective.

Embeddedness becomes even more critical in an increasingly knowledge-based economy where innovation is an interactive process, based on collaboration between a large policy network that can include private firms, government agencies, multiple levels of government, universities, and customers (Evans 2008; Lundvall 1992). Private firms are more reliant on collaborations and knowledge sharing to keep up with fast-moving technological frontiers and market changes; and governments must continuously absorb knowledge from a variety of partners to successfully design and implement policies (see Cooke and Morgan 1998). Governments cannot play their unique roles within these policy networks without being embedded within them.

In a more knowledge-based economy with a need for larger information requirements and co-implementation, Evans (2008) highlights the need to expand the notion of embeddedness from a concept principally concerned with government–business relations toward also incorporating civil society. Civil-society groups can help identify and manage social barriers to technology adoption (e.g., how to promote low-income inclusion in sustainable energy), promote political legitimacy, and help policy initiatives stay on mission. This might require gathering constituencies that are more numerous and less organized. As Evans (2008) states, the twenty-first century calls for “more complex and demanding forms of embeddedness.”

To understand the multiple innovation players that might need to interact to develop a technology, consider the vision of a “smart grid” (Fox-Penner 2010). Successful smart-grid development likely requires the involvement of utility companies, information-technology providers, and private-sector companies developing energy-management systems and digital applications. Governments will need to change regulations (e.g., time of use pricing) and help install new infrastructures. It is also critical that users become directly engaged because many smart-grid technologies aim to change consumer habits (Verbong, Beemsterboer, and Sengers 2013). Users can also provide critical feedback to improve software technologies. Governments will not be able to understand how best to facilitate smart-grid development without obtaining knowledge from each of these groups.

The second section of this paper also noted the danger of governments becoming captured by groups in society, such as a particular industry lobby. An embedded public sector could be more prone to capture. As discussed above, attempting to insulate the public sector is unrealistic: it would cut off information flow needed to create effective policies and would significantly decrease the public sector's capacity to shape innovation trajectories through strategic leadership and coordination.

Institutional design can guard against capture by vested interests by promoting the autonomy of public sector organizations and personnel. In his study of the computer industry, Evans (1995) emphasizes the importance of state autonomy alongside embeddedness. He finds that the most successful states had the right mix of “embedded autonomy” to make the state both more competent with respect to their policy arena and capable of avoiding capture or corruption. Governments engage with private-sector partners, but they do so on their own terms.

Autonomy is achieved by ensuring that those engaged in policy implementation have a high level of expertise and an ability to develop their own visions of technological futures. It can be promoted by following basic public-administration principles such as merit-based hiring. A clear public-purpose mission also helps set the rules for public–private collaborations. The idea is to collaborate on achieving the mission rather than having government work in the interests of a particular client or group. Schmitz et al. (2013) call for “transformative alliances,” highlighting that actors with different interests and objectives can nevertheless pull in the same direction.

The public sector must demonstrate competence in its policy area to support its autonomy and earn trust within its policy network. Leadership, culture, talent, and institutional fit are all factors that relate to the competence of the public sector.

Narayanamurti et al. (2009) highlight the importance of leadership and a culture that promotes integrity and public purpose. They call for leaders with managerial as well as scientific/technical credentials. In the energy field, they highlight the need to understand the role of public organizations in the overall system, with capabilities to “integrate the different activities within and outside the institution.”

The public sector must also attract talented individuals with deep “domain specific competencies” (Jacobsson and Bergek 2011). A mobilizing mission can act as a magnet for talented and motivated individuals (Narayanamurti et al. 2009; Mazzucato 2013). The method of finding and attracting this talent might depend on individual contexts. The US Defense Advanced Research Projects Agency's (DARPA) innovation model makes uses of temporary project teams. Former directors Regina Dugan and Gabriel Kaigham (2013) argue that finite timelines attract high-calibre individuals interested in working on problems but not on becoming permanent employees. In contrast, permanent groups with specialized knowledge of particular sectors are used in Finland's Tekes agency. The temporary projects model might fit the American context where a large pool of talent is available from industry and academia. DARPA also works on producing breakthroughs that connect basic science with user needs, which lends itself to short, focused projects. In contrast, jurisdictions that need to build new capabilities might opt for developing specializations in-house over longer periods. More permanent project teams might also be more relevant to promoting diffusion and conducting innovation system analyses.

The literature suggests that the competency of public institutions should be prioritized over theoretically ideal policy instruments (Lipsey and Carlaw 1998; Rodrik 2007). The choice of policy instruments should be determined primarily by how effectively they can be wielded in practice. As Rodrik (2007) states, “it is better to employ a second-best instrument effectively than to use first-best instruments badly.” Assessing what institutional competencies exist, where they are located in the public sector, and what competencies can be built up in the public sector or acquired by joining up with non-governmental organizations is an important aspect of governing innovation. The possible institutional fits within a given jurisdiction can help determine where policy functions are best placed.

The level of government (federal, provincial, municipal) best placed to play certain innovation policy roles is another aspect of promoting institutional competency. This issue is especially relevant in the Canadian context, given highly differentiated regional economies and federalist political structures (Simeon 1979; Jenkin 1983). Innovation theories emphasize the importance of interactive learning, collaborative policy networks, and geographically specific concentrations of specialized capabilities and skills (see Morgan 2007; Wolfe and Lucas 2005). Local and regional levels of government are often best placed to foster meaningful collaborations and develop strategic priorities. However, these local innovation initiatives must be able to access resources and pull down policy supports and intelligence from upper levels of government (see Creutzberg 2011; Bradford 2004; OECD 2014). The federal, provincial, and local levels of government have separate institutional competencies that need to work in concert.

To gain trust within networks and to send the right signals to the private sector, the public sector must have the ability to do what it says it will do (Schmitz et al. 2013). Private firms that change business strategies based on public policy expose themselves to investment risks in the case of unexpected policy change. The private sector must have confidence that the public sector's policy directions are credible and will not create unexpected changes. An unpredictable environment can stymie an otherwise robust policy framework. In Ontario, Holburn, Lui, and Morand (2010) found that policy uncertainty was the most important factor in wind-firm location decisions, and that uncertainty led to higher wind-power prices.

The development of innovation often requires much longer periods than are typically resourced through government funding cycles. A boom-and-bust funding cycle can disrupt long and complex technological trajectories (Khanberg and Joshi 2012; Narayanamurti et al. 2009; Foray, Mowery, and Nelson 2012) and can result in the loss of talent in public-sector organizations as employees look for work elsewhere. This in turn reduces credibility and trust with private-sector partners.

Authors making recommendations for significant increases in clean-innovation research and development activities have also called for steady changes in annual budgets (see Dechezleprêtre, Martin, and Bassi 2016; Dechezleprêtre and Popp 2015; Newell 2008). As demonstrated by Freeman and Van Reenen's (2009) study on the US National Institutes of Health (NIH), large increases and decreases in R&D budgets can create significant adjustment problems. In the face of rapid budget increases, the NIH needed to rapidly attract researchers, who needed time to develop their knowledge and expertise. Then a rapid decrease created career problems for these researchers who needed to compete for a smaller pool of research funds.

Clean-innovation policy must be designed and implemented in the public interest. Accountability is required to guard against political capture or self-interested behaviour on the part of agencies. Public organizations have a responsibility to “explain what they are doing and how they are doing it” (Rodrik 2014). This demands a high degree of transparency from the outset. Rodrik (2007) calls for a clear link with a political leader who is accountable for the consequences of innovation policy and capable of monitoring performance. He also notes that a political master should not only engage in oversight but also be a champion with a high level of authority. This ensures that the issues related to economic transformation have a strong voice in government—equivalent to the role of the finance minister in promoting fiscal prudence.

A crucially important yet difficult question concerns the evaluation of clean-innovation performance. Evaluation can maintain accountability, build political legitimacy, and promote policy learning. However, if done badly, evaluation can dramatically increase overhead, impede flexibility, and reduce comprehensiveness by narrowing activities to those that can most easily be evaluated rather than those that could be transformative in nature (see Savoie 2015).

There is no single metric to evaluate innovation performance. Restricting evaluation to a limited number of indicators (e.g., patents) will reveal only limited aspects of the innovation picture and could misdirect policy-makers toward supporting only one aspect of the innovation system. A multitude of indicators can be used to monitor innovation, such as the number of new products and firms, quality improvements, and price performance. In addition to monitoring output indicators, policy-maker actions are often best informed by tracking innovation activities or processes such as network-building, market formation, entrepreneurial experimentation, and initiatives to increase human capital (Edquist 2004; Hekkert and Negro 2009; Bergek et al. 2008). To maintain focus on a clean-innovation mission, environmental indicators should be tracked and given significant weight in the evaluation of projects.

The other important aspect of evaluating innovation is that the failure of some projects is inevitable, since innovation is a process of learning and experimentation. An institutional structure should ensure that technology successes are able to partially offset the inevitable losses, and therefore institutional evaluation should be based on the entire portfolio of “winning” and “losing” projects over a sufficiently long period. In addition, project success should be judged based on social or systemic objectives such as the creation of spillovers across the economy, which can occur in projects that both succeed and fail to reach commercialization stages (McDowall and Ekins 2014; Rodrik 2014; Mazzucato 2015). The key is to ensure that policy-makers learn from failures. In their evaluation of innovation policies, Lipsey and Carlaw (1996, 1998) list two types of successes. The first type leads toward the successful commercialization of a process or product; The second type of success is a situation where an attempt was worth making at the outset, the attempt fails, and the failure is recognized in a timely fashion and support is terminated.

Learning from both successes and failures requires government policy-makers to have a high capacity for self-monitoring or “diagnostic monitoring” (Sabel 1994; Kuznetsov and Sabel 2014). Evaluation frameworks that support iterative learning and consider course corrections could be highly valuable (Arnold 2004). These types of evaluation frameworks might not easily fit within more standard cost-benefit frameworks, but they might be the best way to support the public sector's unique role in promoting low-carbon innovation.

How to appropriately evaluate clean-innovation policy in the Canadian context deserves careful consideration from policy-makers, and this issue requires further research. The challenges of properly evaluating innovation should be discussed upfront to avoid misunderstandings that lead to unnecessary controversies.