Canada’s New Innovation and Investment Agency: How data could help foster innovation
by Nicolas Sacchetti
Data driven enterprise empowerment – Helping firms help themselves to innovate.
Catherine Beaudry, director of 4POINT0, led a discussion on May 10, 2022, about the Indicators and Data Effectiveness Assessment to Foster Impactful Innovation (IDEA) Hackathon results and ways forward to a Statistic Act 2.0.
The panel was composed of:
Guillaume Paré – Advisor to the Director of Research and Innovation, Polytechnique Montréal;
Jan Kestle – Member of the Canadian Statistics Advisory Council (CSAC);
Loick-Alexandre Gauthier – Senior Director of the Institut de l’innovation du Québec;
André Mouton – Executive Director International at the National Research Council of Canada (NRC);
Pierre Mohnen – Professorial Fellow at UNU-MERIT and Maastricht University
Design – How could the design of business innovation support programs be modified to include data capture to facilitate delivery and assessment?
Delivery – How could we use available data to improve the effectiveness of government supports available?
Assessment – How could we use available methods or indicators that address the challenges of evaluating a business support program, such as the existence of multiple or successive supports used, or multiple targeted outcomes?
This is a summary of the discussion which took place at the 2nd International P4IE Measuring Metrics that Matter Conference on Policies, Processes and Practices for Performance of Innovation Ecosystems.
Prior to what has been said, here are the IDEA Hackathon questions and two teams’ proposal to examine how data and indicators could be used or captured to better inform government initiatives:
Team 1 proposal goes as follows:
« The formation of a “Precision BIGS” overseen by the new Canadian Innovation and Investment Agency (CIIA) to provide support and personal recommendations to firms. The recommendations would be informed by Statistics Canada. Prior to what has been said, who would collect data in the same way it did with the original Business Innovation and Growth Support (BIGS) database. The recommendations would additionally create personalised benchmarking reports to the various firms that have answered. It is a free consultation/benchmark structure to funnel firms to the right institutions for funding/support that doubles as an incentive for them to provide data to Statistics Canada. »
Team 5 proposal goes as follows:
« A centralized customer relationship management (CRM) portal/database that shares basic information (including company data, programs/service data and departmental data) about the recipient company of a given government program/service between departments. Public servants working with recipient companies can use the CRM portal to determine which departments have already served this company and therefore coordinate delivery of other services. Specific program deliverers can also use the portal to seek out potential new recipients/clients. »
On the key takeaways of the executive panel made of 4 senior management level judges, it has been mentioned that « there may also be an opportunity to integrate Team 5’s CRM system into Team 1’s “precision BIGS” consultation/benchmark structure. »
Discussion
Catherine Beaudry: André, what were your impressions of the IDEA Hackathon results?
André Mouton: I was particularly impressed by Team 5’s proposal. As part of the NRC for the Québec region of the Industrial Research Assistance Program (IRAP), we are trying to have SMEs across Canada to grow in technological field. Having information about them is difficult. Access to data for us means awareness of what is going on in the region and whatever high potential SMEs IRAP could have. We have a data vision that one day we can link companies across Canada based on interests, synergies.
Team 5 is more or less proposing to have some ways of grouping information. It is a great idea and I hope we can move forward with that. However, we already have some data, but it is difficult to find, manipulate and use. At least, we need to have some ways of being able to have some ideas of companies by region.
The information on the registry is very non-homogenous across Canada, depending on regions and provinces. Sometimes it is very bare minimum and close to impossible to do anything. I think you will have an enormous regional development if we could have access to this information and I see StatCan as being one of the institutions that can help a lot in this moving forward.
Loick-Alexandre Gauthier: What hit me was the need for data centralization. All teams mentioned the aspect of how we can create value with data. If we want people to trust Statistics Canada, data have to give them information to help them make decisions. Plan actions, for example, tackle climate or social impacts.
Pierre Mohnen: First, researchers would like to have representative data. Selection bias is something we would like to avoid. The second one is to have panel data. It means to have observations overtime over the same firm, or the same individual, as long as possible so that we can control for an observed heterogeneity – things that we cannot measure. If we have panel data, we can control for it. We can also look at the dynamics: a decision today can take one or two years before you see its effects in the data, and for that you need panel data.
Of course, we would like to have updated data – although it may be less important for the researcher than for the decision maker – not to have 6-7 years old surveys. Then, especially important and mentioned in the hackathon is to link different datasets: the innovation survey data, the labour force data, investment data, trade data, production data. All this can be integrated. Then you can really have an interesting model with all kinds of explanatory variables to explain why some innovations are more important for the Canadian economy than others.
The next quality would be comparability. It would be so nice to be able to compare the effectiveness of tax incentives or the effectiveness of subsidies in support of innovation across different countries. But this international comparability is still nowadays very difficult to do. The OECD has been trying to execute these things with innovation survey data. The same model was estimated with data from various countries, but no one had access to all the datasets. You could only work with your own national datasets.
Every national statistical agency had its own innovation survey. Those surveys were not necessarily conducted in the same way. For instance, in one country it would be a sample of enterprises above 20 employees, in another it was above 50 employees, and so on. That makes international comparability difficult.
As academic researchers, we would like to have access to as much data as possible. But, of course, it is not always reasonable to ask Statistics Canada to do all kinds of surveys. As one of the hackathon proposals, it would be interesting to centralize the different data and to homogenize it. That is not easy because different surveys use different samples. When you put them together, you lose a lot of observations; firms that are not in the same samples. These are difficult things, but Statistics Canada is doing an excellent job at this.
The problem of confidentiality is, of course, important. But it would be interesting to ask firms: after five years, are you ready to release your data? I would not be surprised to hear that firms are willing to release five years’ old data.
The last point is to evaluate the effectiveness of these different subsidy programs. From the beginning, the program should be devised in such a way that afterwards we have the data to check whether it was effective. A deal could be made with the respondents: we give you some subsidies, but in return we want you to also give us some of your data.
Jan Kestle: Well, I always say you cannot change what you do not measure. So, I would reinforce that last comment. When whatever we are doing to offer support and strategy for innovation and growth to SMEs, we must have a better way of measuring outcomes and designing the research up front.
I know the problem of mixing firm data from one study to another and the representative samples are really challenging. But there are a lot of synthetic methods: rolling panels where you can come up with one person or one firm to replace another.
It comes down to leadership. The federal government must empower data strategy, integrity, and digital charter. It is not just about technology of data and information, it is about the methodology, the quality and ability for researchers to do these things.
Most of the challenges identified by the three analysts (Mouton – Gauthier – Mohnen) really come back to funding and leadership. You do not solve the federal/provincial and the reconciliation problems easily, but we cannot have those things prevent us from moving forward with good research. Calling on the federal government to provide more authority and resources to lead in research and innovation in terms of core statistics and informations we need.
Catherine Beaudry: I think there is a key point to stress. The three of you [Mouton – Gauthier – Mohnen] have quite dissimilar needs in terms of data. Yet, the source is the same. For an organization like Statistics Canada, pleasing all three of you may actually be quite complicated. Pierre and Loick, you do not need to know the name of the firm, where it is from, or even have recent data. But for an organization like IRAP, if you are to deliver programs, funding, help, support, mentorship to a firm then you need to know who that firm is. So, you need to collect other data.
André Mouton: What you said is so true. We are just working in the company. We are not interested in the people working for the company. We were thinking it would be easy to have access to data activity in terms of technology and market. It is not the case at all.
Plus, the multiplicity of names that companies use in different registry makes it impossible to know which company is which. Is it French, English, commercial or legal. We were thinking the information would be available, but it is not the case. It is quite different from playing with people and having to face personal data.
Loick-Alexandre Gauthier: For the Québec government, just to see how policy impacts companies in general, if we use the data from StatCan it could be difficult for us to have specific or regional data and to take decisions. We want to see the world, for example an industrial sector, how it is performing compared to other sectors either from Ontario or B-C for example.
So far it is much more about comparing Canadian regions, even other countries and companies’ performance here in Québec. And are our policies adequate responses to companies in general, or do we need to make changes to improve?
Pierre Mohnen: What we, researchers, need is to link different datasets. For instance, employer-employee data or data of the innovation surveys linked with patent statistics.
Catherine Beaudry: Did you find solutions to some of the challenges that you mentioned in some of the hackathon proposals that were provided?
Loick-Alexandre Gauthier: The need for data centralization and to ease access to programs and subsidies to companies caught my attention. Right now in Québec there are quite a lot of programs, from the provincial and federal. But it is generally difficult for a company to get help. For every program, you must fill in a different form.
Homogeneity of what is asked for companies is key to easy access. This way you will have data homogenization and it would be much easier for researchers and on government level to use that data. Right now, you need to go through different ministry levels and ask to get access to specific data about companies or programs and every time it is different.
André Mouton: I completely agree with what Loick just said, for us it would be important, because of the homogeneity of data and the ability to link diverse sources.
Catherine Beaudry: Jan and Guillaume, we heard you this morning talking about data privacy and firms having a say about what data is shared, or the triangulation – you can get more information on firms or an individual if you are collecting many data points from many data sources. You are probably collecting more information about these firms than the firm or the individual knows. Can you both comment on that? Jan from a private sector point of view and Guillaume from an ethics point of view?
Guillaume Paré: The coupling of data always raised the issue because it reveals things that even the person that provides the data did not know about themselves. That is all the issue of social media and the bigdata usage. However, I think this issue can be counterbalanced by a shared community of not only meaning but values of data usage.
For example, in the academic world most of the time it is the quest for knowledge that will drive the researchers to be able to assess a situation, criticize something or reveal the truth on a particular phenomenon. These are all led by different values shared most of the time by our government and our society. We must have that kind of symmetry in the values within the usage of the data.
Jan Kestle: Whether we are talking about privacy for individuals or confidentiality for firms, we need to start from a principle-based approach. We know legislation needs to be changed in the Statistic Act and in the privacy laws, but the reality is that most organizations strive to do the right thing.
From my experience, I think that most researchers, data analytics firms, banks and telecom who collect data from individuals, are trying to provide better products for better services, trying to make people’s lives easier. The small number of bad actors that we get in the headlines every day distorts the narrative.
Since 1997, Canada has been living in a strict privacy environment. Private sector businesses, government and NPOs have spent millions of dollars to understand the laws and to obey them. There will always be breaches and exceptions but people who believe in evidence-driven decision-making – whether it is in the academic community or the private sector – are trying to do the right thing.
Part of the reason we are all faced with this right now is that the external world can create data and use data exchange so much. It is not a business-as-usual environment. When we talk about centralization, we need an organization to create those containers and standards on the metadata to understand the needs of three distinct kinds of users that are assembled there.
We must start not by designing the perfect solution at the other end but by building from the ground up, taking the data that we have and organizing and assembling it in a way that you can meet the needs of different users.
A lot of the data comes from Statistics Canada. It is collected in the same way, but if this data is assembled according to international and local standards around the linkages, we can build something better.
I have been working on the data field in Canada for five decades. During these decades, the biggest issue is we need a better, more accessible, and consistent business register. We need the federal/provincial problem solved. We need the firm versus establishments solved.
You cannot get perfect but there are tons of technologies that can help with matching, cleaning, and standardization. We need to invest in a Canada who wants to be data-driven. To be innovative we need the federal government because that’s where the responsibilities start. We need the federal government to invest in better data resources, centralized resources, standards, processes, and stakeholder alignments so we can move this forward.
This content has been updated on 2023-10-27 at 21 h 36 min.