As the environmental benefits of technology such as solar panels and electric cars become more widely known, both local and national government officials are exploring policy solutions to encourage consumer adoption.
But these polices, which often include subsidies to achieve a certain level of consumer adoption by a target date, often don’t take into consideration the ripple effects such programs create, such as word-of-mouth or the increased efficiency that comes as more and more of the technology is brought on line. In “Consumer Choice Model for Forecasting Demand and Designing Incentives for Solar Technology,” Wharton operations and information management professor Ruben Lobel and co-author Georgia Perakis of MIT’s Sloan School of Management present a framework that policy makers can use to create an optimal plan for offering and then phasing out these incentives.
In this interview, Lobel describes their findings and also how the researchers applied them to studying the efficiency of the solar market in Germany.
An edited transcript of the interview appears below.
Knowledge@Wharton: Can you give us a short summary of your research?
Ruben Lobel: The research we are trying to do is to give an operational perspective to policy making, in particular for the adoption of solar panels…. It’s common to see subsidies that will help green technologies take off faster. These technologies are very expensive in the beginning, and they need some sort of public support in order to become economically viable. We see a lot of policies out there that have a certain target in mind: In Europe, you had the goal of getting 20% of energy from renewable sources by 2020; in California, it was 33%. You had the million rooftop programs that wanted to cover one million rooftops with solar panels in Germany.
And this is the inspiration we started from. Instead of trying to understand where these policies come from, we were trying to understand, given the policy with a certain target, how do you get there in the most efficient way, and in the cheapest way for the taxpayers? We found that it’s important to consider the feedback loops in these market dynamics: As you have more solar panels being adopted, that makes it cheaper for the next solar panel to be sold. On average, that makes for more word-of-mouth being spread about that technology, and a good policy-making practice should incorporate these feedback loops when designing the policy.
“We were trying to understand, given the policy with a certain target, how do you get there in the most efficient way, and in the cheapest way for the taxpayers?”
Knowledge@Wharton: What were the key takeaways from this research?
Lobel: The key takeaway from this research is that you should take into account these feedback loops, such as learning by doing, and information diffusion when designing a policy. The outcome of the optimal policy designed to achieve a certain target is not trivial, in the sense that you should not just make a lump-sum investment in the beginning and let the market figure itself out. You actually can observe a certain pattern. Yes, early subsidies do stimulate adoption. But they also impact future subsidies. So there is a certain way you should distribute subsidies throughout the time horizon in order to reach your target in the most efficient way.
Knowledge@Wharton: As you worked on this research, what surprised you the most?
Lobel: A typical economics approach to understanding policy making starts with a definition of system welfare. And we abstracted a bit from that. We went to the next level of decision making. Suppose that I don’t know what the welfare is, but I have a given target. I assume that this is the target of adoption that I need to get to. It’s hopefully motivated by some kind of welfare optimization — i.e., that society will be better off by some objective measure [as a result of whatever policy is being implemented]. When we develop this framework, what we try to do is to work out what would be the optimal welfare function that would justify [setting a particular subsidy policy].
Historically, we see in Germany a certain level of subsidies being given out. And I’d like to try to understand what sort of economic welfare measure would justify that given historical path. We were surprised to see that there is no such welfare function that would justify that history. There is a way to show, just using this framework, that [to reach the target more efficiently], more subsidies should have been required in the beginning — a stronger subsidy policy, perhaps — and a stronger phase-out in the later stages of the program. Those are the indications that we have.
We have to put a qualifier on it. We don’t do a full empirical study of the German solar market. We have very limited access to data. We just use aggregate data for the yearly adoptions and the average levels of subsidy. But it’s a small indication that we can try to measure the efficiency of a policy program and characterize it as being somewhat inefficient, even without having a clear measure of welfare.
Knowledge@Wharton: What would you say are the key practical implications, particularly for policy makers?
Lobel: The most practical implication is to try not to be myopic. That’s the most important recommendation. Try to understand the future implications of the policies you are making today, and the market dynamics. A lot of policy sometimes is decided by [asking], “What is the level of the subsidy that we are going to do? Let’s try to give the investor a certain rate of return, and we’ll hope that gets us close to our target.” But when you do this over several years, you have to understand that early adopters create an impact on future adoption. And if you don’t take that into account, then you’ll either miss your target or you’re going to spend a lot more money than you should.
“[Early] adopters create an impact on future adoption. And if you don’t take that into account, then you’ll either miss your target, or you’re going to spend a lot more money than you should.”
Knowledge@Wharton: Can you tell me about how the research applies to makers of solar panels or people who might want to buy them?
Lobel: Companies need to understand how learning happens in their industry and how it will affect future policy decisions. One of the most problematic things we have in this area is the abrupt changes in subsidy policies. That has something to do with some of the future research that I’m working on, trying to understand how policy making affects the industry, and vice versa. There’s a strategic interaction between how companies are behaving and how policy makers will respond, and vice versa.
For consumers, I think it’s important for them to try to understand how to time the market. From anecdotal evidence we’ve seen, a lot of consumers are trying to wait for solar panels to be cheaper. They’re not necessarily doing the optimal thing. From what we have observed, the rate of subsidies has been decreasing faster than the cost of the installation. So in a way, timing yourself in this market is not necessarily as easy as you might think.
Knowledge@Wharton: Can the key findings that you had here apply beyond the market and subsidies for solar panels? Can it apply to other green technology, or other industries as well?
Lobel: Absolutely. We are doing a few other studies in the areas of smart grid applications and electric vehicles. There is a small study where we use the adoption of electric cars. And it’s a very similar type of market dynamics that are happening. The same lessons — that you need to incorporate learning by doing, information diffusion — are also important to be considered in these situations. You can also think about companies trying to acquire market share in certain products. They will have to think about it in a similar framework as we do in this paper. Some of the extrapolation from this paper can go into other areas.
Knowledge@Wharton: Is there any story in the news that’s relevant to the research?
“There’s a strategic interaction between how companies are behaving and how policy makers will respond, and vice versa.”
Lobel: The actual subsidy program in Germany has taken a lot of flak in recent years for overspending at a time of [economic] crisis. And that’s something that needs to be taken with a grain of salt. They have developed a huge industry of solar installation. They’ve driven down the costs of installations much further than places like California, for example.
If you compare how they’ve done it, it was through strong, timely interventions in the beginning. And in this paper, I talk about how it could perhaps have been even faster [using stronger subsidies] in the beginning, and slowing later on. They’ve only started to do a more aggressive phase-out in the last few years. Although they had a planned phase-out, they could have done it a little bit faster and avoided some of this criticism, perhaps.
Now, the main criticism is that these solar panels that customers bought became too profitable for them. And all the other rate payers in the country are going to be paying a huge penalty for the adoption of these solar panels. It’s hard to quantify how successful a program is. It’s hard to quantify what the other benefits are that this industry that emerged in Germany will have.
An example of a country that has a not as successful program was Spain, which started about the same time and was very aggressive in the market, perhaps in some cases even more aggressive than Germany. They had more solar radiation than Germany, so potentially it would be more efficient to install panels in Spain. Yet the industry was growing too fast and without developing local manufacturing and local technology. The government backed out and brought the entire industry to a near collapse. That’s a clear lesson of how not to do it, as opposed to what Germany did.
Knowledge@Wharton: What are you planning on looking at next? Can you talk about some of the research that follows this particular paper?
Lobel: In this paper, we talk about these feedback loops like learning by doing and information diffusion. And in the next few papers that we’re working on, we explore a few different parts of the market. For example, we look at how demand uncertainty will affect the subsidy programs. We look at how policy flexibility versus commitment will affect the industry. For example, say you’re a policy maker and you have a certain target, but are given a deadline [for achieving it.] Would you like to readjust your policy as you approach the target deadline to make sure that you hit your target? Or would you like to commit to a longer term policy?
What we show in that situation is that you would rather commit to a longer term policy. With commitment, you do lose some flexibility in the long run, but you give a signal to the people in the market that they need to put in their fair share and they need to absorb some of the risk. When you’re a policy maker with too much flexibility and you readjust your policy very often, you send a bad signal to the industry. So that’s the future direction of the research that we’re currently working on.