Firm organization - tradeoffs in make vs. buy

Hi, welcome back. In this video, we will talk  about firm organization. Specifically we will think about make or buy.

We will shed light on the pains and gains of collaboration in traditional industries, as well as in dynamic digital environments. We will be thinking a lot about various trade-offs in finding the optimal organizational form. We will see that we can model trade-offs as non-linear relationships  between inputs and outputs - as functions that have a maximum somewhere.

Applying this logic to a regression framework, we can use data to try to estimate where the optimum is. If this makes you uncomfortable, you should take a look at the econometrics refresher videos. If this sounds exciting to you, stay tuned and I will show you some evidence from papers along the way.

First, let’s think about the pros and cons of collaboration. Perhaps the biggest reason why firms choose to collaborate is to achieve  cost and quality advantages due to specialization. Collaboration can be a substitute for a lack of internal capabilities.

But you also don’t want to outsource too much. Why? Because there are often complementarities between internal and external knowledge.

One way to think about this  is absorptive capacity. When firms invest in own R&D, they are better equipped to learn from other firms. That is, even if there are incoming spillovers, it can make sense to have an own R&D unit because investing in R&D produces the ability to absorb information from  the external environment.

But of course, building up your absorptive capacity is not free. Part of the reason is that knowledge from the external environment will have to be adapted to  be fully useful internally. Hence, there needs to be  firm-specific redeployment; someone needs to be in-house to figure out how  to integrate the externally created knowledge into the organization.

There are also other problems in collaborations. Perhaps most importantly, there is the issue of how to allocate  intellectual property rights. Imagine you contract a firm; you outsource your R&D to this firm, and they come up with something that shows  great potential to be commercialized.

How do you make sure that they  do not runaway with this idea? Maybe you can write a contract to define property rights ex-ante, but this could incentivize the  outsourcing firm to lower their effort. Another issue are outgoing spillovers that allow your competitors to learn from you.

Imagine there is a vendor that has a central position between you and your competitor. It might be that whatever the vendor is doing for you, they are also doing for your competitors. Maybe your competitor is working on a similar sets of problems and the solution that is useful to  you is also useful to your competitor.

Information asymmetry is also often a problem in collaborations. If your partner is too specialized, you might not fully understand what they are doing and it is difficult for you to assess the quality of their work. The gains and pains of  collaboration or outsourcing can be best summarized in a simple graph.

On the x-axis, we plot the degree of collaboration and on the y axis we’re looking  at some measure of performance. Collaborating more is improving performance, because you benefit from the gains. But then at some point you start to see the countervailing forces more and more.

The gains of collaboration are getting more important. You arrive at a curve that  looks like an inverted U. Now let’s look at a paper that does exactly what we have discussed.

Kaiser and Grimpe study the relationship between R&D outsourcing and innovation performance, and want to see whether there is an inverse U-shape. They have firm-level data that covers a broad range of industries. They estimate a model where the dependent variable is the share of sales of new products.

R&D outsourcing is measured as expenditure for external R&D again as share of sales. Let’s look at some of the results. In the first model they just used a linear term, so we see that there is a positive relationship; more outsourcing leads to higher innovation performance.

In models 2, 3 and 4, they introduced the square term. So now we can compute the extreme point - and we arrive at 1. 157.

Remember that R&D outsourcing here is measured as a share of sales, so it does not make much sense that firms invest more in R&D outsourcing, then they have sales. We would not expect a number greater than 1 to make much sense. Thus, if we take the coefficients seriously it would imply that this is more a case of diminishing returns.

If we do the same computations in the other models we arrive at more reasonable extreme points. It all depends on what other control variables the authors have considered. The results look like that they support the ideas of pains and gains of collaboration.

Grimpe and Kaiser go one step further. They show that you can move the extreme point to the right – that is, you can benefit more from more outsourcing - if you invest in internal R&D or collaborate more intensively with external firms. Let me summarize.

In this video we talked about the pros and cons of collaboration in dynamic and less dynamic settings. We have also seen how we can know whether a non-linear function is U-shaped or not through simple computations. And finally, we looked at various papers that confirmed our theoretical inputs.