How can Sweden pursue a successful innovation policy?

– A methodological report on comparing results from political interventions

This report takes an important step toward developing and rethinking innovation policy-based on evidence. The state invests in a wide range of initiatives in this area, from financial support for young, innovative tech firms to research centers that promote collaboration between academia and industry.

The aim of the report is to provide better insight into the outcomes of these efforts. This enables a comparison between the costs and results of different initiatives—for example, how many patents or research positions are generated per krona spent, and which types of support are more effective than others.

This approach has been used for years in development economics. It has led to a structured way of generating evidence to inform policy decisions—and raised the level of knowledge in both development economics and related policy areas. A clear example is the many studies that examine which interventions most effectively improve children's literacy. Such comparisons show which measures help the most children learn to read within a given budget. We see strong potential to apply a similar approach to innovation policy. There are methodological challenges, and the work will take time—but Sweden has the opportunity to lead the way.

The report conducts two analyses aimed at explaining and testing how different innovation policy interventions can best be compared. First, it identifies and discusses methodological options for analyzing the costs and effects of such interventions, drawing on the existing research literature. Second, it presents a cost-effect analysis of Eurostars, a direct support program for research and development, based on insights from the literature review. Eurostars has previously been evaluated by Growth Analysis and is used here as a case study.

Choice of method for comparing interventions

Cost-effect ratios are useful

We have examined which method is best suited for comparing innovation policy initiatives. Our conclusion is that cost-effect ratios, specifically the "value-for-money" (VfM) approach, provide a strong starting point. This method links costs to achieved outcomes, such as how much firms increase their R&D investments per public krona invested. A key strength of the method is that it builds on impact evaluations to estimate the size of the effects, and it also allows for the inclusion of outcomes that occur at different points in time.

The literature includes examples where variants of VfM ratios, such as cost per job or the increase in R&D spending per krona of support, have been calculated for innovation policy initiatives. However, the methods used often lack transparency and consistency, which makes comparative analysis more difficult. In this report, we have focused on measures targeting firms with the aim of promoting innovation, but the method can also be applied to firm-oriented initiatives with broader objectives — such as sustainability or national security — provided that impact evaluations are available for the respective outcome variables.

The literature review shows that cost-benefit analyses have been used to assess welfare effects, but the number of such studies is limited. It is also methodologically demanding to model both innovation outcomes and costs in a reliable way. Despite these challenges, our review suggests that cost-benefit analysis holds potential as a method. The key is to adapt it to better capture the varied effects of innovation on firms and the broader economy.

The conditions for comparing policy interventions need to be improved

A basic requirement for comparing different interventions using the VfM approach is access to reliable impact evaluations and the systematic compilation of cost data. In recent years, the reliability of impact evaluations has improved as new methods have become more widely adopted, leading to a deeper understanding of how innovation policy measures affect firms' innovation capacity and performance.

At the same time, there is still a lack of impact evaluations based on Swedish data. It also remains uncommon for intervention costs to be documented and compiled systematically, which leaves the overall picture of cost-effectiveness incomplete. Finally, there are methodological challenges in linking costs to effects, especially when assessing societal benefits and net impacts.

Case Study – The Eurostars Programme

We tested the VfM method on an actual policy initiative: Eurostars, a support program for small and medium-sized enterprises engaged in research and development. Using previous impact evaluations and new cost data, we calculated several VfM ratios. The results show, among other things, that:

• Stimulus to private R&D investment. For every krona in public support, firms increased their own R&D spending by an average of 5.3 kronor. This suggests that the state support acted as a catalyst, prompting firms to invest significantly more in innovation.
• Job creation focused on innovation. The 243 firms that received support between 2008 and 2019 created a total of 831 new jobs. Of these, 71 percent were in research and development, indicating that the support led not only to more jobs but also to more high-skilled positions that enhance firms' innovation capacity.
• Increased turnover — a sign of commercialization. The firms' turnover increased by a total of SEK 2.6 billion over seven years following the support. For each krona the support cost it generated on average an increase in turnover by 3,97 krona, a measure reflecting business development and the commercialization of new goods or services.

To calculate the VfM ratio, we not only assessed the outcomes of the program but also analyzed its actual costs. The costs were divided into four components:

• 74% – Direct budget outlays. The largest share was the support paid directly to firms. This represents the public funds the state invested in the program.
• 9% – Administrative costs for firms. Supported firms incurred their own costs in preparing applications, reporting results, and managing projects.
• 2% – Government administrative costs. A small share of the total cost was incurred by public agencies administering the program, such as reviewing applications and monitoring implementation.
• 15% – Cost of public funds. Using tax revenue entails indirect costs to society, such as tax collection burdens or opportunity costs. This is often referred to as the cost of public funds.

Overall, the analysis shows that Eurostars performs well compared to similar programs, especially in stimulating private R&D investment. However, the report also highlights that comparative analysis remain methodologically challenging due to factors such as:

• Different studies use varying outcome measures and time horizons
• Information on the actual costs of interventions is often lacking

Our recommendations

• The government should work towards the systematic use of VfM (Value for Money) analyses in innovation policy. This is because more analyses provide better foundations for policy development and reassessment. The goal is to make different types of innovation policy measures comparable.
  • Growth Analysis continues to produce VfM analyses and develop a method for comparing these with its own and international studies. The results are compiled into policy briefs for the government.
  • The government should require implementing agencies to compile and report costs.
  • The government should continue promoting more impact evaluations (see Growth Analysis 2023a), preferably already in the planning phase of new initiatives – as is done in Swedish labour market policy (see Growth Analysis 2024a).
• The government should assign a relevant agency, such as the Swedish National Financial Management Authority/the new Swedish Agency for Public Management or Growth Analysis, to develop common guidelines on which effects to measure, how to value them, and over what time period.
• The government and agencies should collaborate internationally, for example within the OECD, to develop common methodologies and comparative data.