An excessive development of green products?

Similarly, Simple Green offers a separate brand, Simple Green Naturals, as “100% naturally derived, with ingredients originating from nature.”

More generally, among all newly introduced products in the US, the percentage that claimed to be green increased from 1.1% in 1986 to 9.5% in 1999; see Kirchhoff (2000).

Green products could in the long run replace brown products. However, the transition between these goods can still take several years, as the example of hybrid cars suggests. For instance, Toyota has simultaneously produced both the Toyota Prius and fossil-fuel cars since 1997.

Solid-recovered fuels provide an example of a good that, despite being relatively green, is still controversial given that its environmental performance is relatively weak; as recognized by the European Recovered Fuel Organization (EN Report 15359). In particular, these fuels are produced by shredding and dehydrating solid waste consisting of largely combustible components of municipal waste. Another highly cited example is that of oil sands, which require an extremely large amount of water for every gallon of oil produced, and that generate more GHGs emissions than standard oil drilling facilities.

Spence (1975) considers a monopolist’s decision to invest in quality, and shows that equilibrium outcomes are not necessarily optimal. While we also demonstrate that the monopolist’s decision to develop a green product can be suboptimal, the parameter conditions under which this case arises shrink as brown and green products become more pollutant. Furthermore, our paper considers firms’ incentives to develop a new line of green products in addition to the existing brown good that firms traditionally produce, thus giving rise to BSEs that do not exist in Spence’s model.

Similar to our paper, they show that firms competing for socially responsible consumers (e.g., consumers with environmental concerns) can lead to an excessive provision of public goods. Arora and Gangopadhyay (1995) consider two firms, each selling a single good and deciding its degree of cleanness and its price.

Their model was extended by Lambertini and Tampieri (2012).

See Dosi and Moretto (2001), Cason and Gangadharan (2002), Mason (2006), Hamilton and Zilberman (2006), Greaker (2006), and Ibanez and Grolleau (2008) for the specific practice of ecolabeling, which is often regarded as CSR.

Their paper has been extended to settings of environmental externalities by Amachera et al. (2004).

For completeness, “Appendix 1” analyzes the case in which both firms simultaneously and independently decide whether to produce a green good.

This demand specification is, thus, similar to that of Singh and Vives (1984) for the analysis of firms’ incentives to compete in either quantities or prices when they produce differentiated products.

More generally, parameter \(\lambda \) captures the product differentiation between the brown and green goods, thus allowing the parameter to embody both the products distinct instrinsic characteristics (e.g., acceleration and cargo space in hybrid and fossil-fuel cars) and their different environmental properties. However, if parameter \(\lambda \) only captures the intrinsic features of the two products, different cases can arise. For instance, when goods are completely differentiated, i.e., \(\lambda =0\), they exhibit totally different intrinsic characteristics, and thus each of them has its own separate market. In this setting, their environmental features can also be completely different (e.g., if the pollution intensity of the green good is zero), similar (if it is the same as that of the brown product), or take intermediate values (if its pollution intensity is smaller).

While the introduction of a green product entails an overall increase in demand, such a development is costly, implying that firm \(i\) does not necessarily find profitable to develop the new product, as we describe in the equilibrium results of Sect. 3.

For compactness, we do not include here the expressions of equilibrium profits in each entry setting. Nevertheless, the proof of Lemma 1 provides them.

While “Appendix 1” analyzes equilibrium development strategies in the simultaneous version of the game, we focus on its sequential version as most real-life examples of firms adding a line of green products to their existing brown goods did it sequentially. For instance, Toyota was the first automaker to offer hybrid cars, the Prius, along with their other (more polluting) cars, in 1997. Other automakers followed by developing their own hybrid cars afterwards: Honda introduced the Insight in 1999, Mitsubishi the Colt in 2005, and Nissan the Leaf in 2010.

Figure 1 considers \(\lambda \le \overline{\lambda }\), where cutoff \(\overline{\lambda }\) becomes \(\overline{\lambda }=2/3\) in this parametric example.

In addition, note that if products are completely differentiated, \(\lambda =0 \), firm 2 develops the green product if \(K_{2}<\frac{(1-z)^{2}}{9} ( K_{2}<\frac{(1-z)^{2}}{4}\)) upon observing that firm 1 developed (did not develop, respectively) the new good. Our equilibrium analysis at the end of this section provides comparative statics of cutoffs \(K^{A}\) and \(K^{B}\).

If firms instead simultaneously choose to develop the green product, the results in Proposition 1 still apply; except for point (3) which holds under different parameter conditions. For more details on the equilibrium of the simultaneous-move version of the game, see “Appendix 1”.

A similar representation applies to the second strategy profile, \( (G_{1},NG_{2})\), depicted in areas (2) for firm 1 (in Fig. 2a) and for firm 2 (in Fig. 2b).

In particular, if only firm \(i\) had the ability to develop the green good, it would do so when development costs are relatively low, \(K<K^{B}\). However, when firm \(j\) also has the ability to develop the green good, firm \( i\) might decide to develop under more expensive development costs \(K<K^{A}\), where \(K^{A}>K^{B}\).

For more details, see http://​www.​babygearlab.​com/​Disposable-Diaper-Reviews/​Huggies-Pure-Natural.

For simplicity, our social welfare function abstracts from the cost of raising public funds. Extended models could consider this cost if the regulator provides subsidies to lower firms’ development costs (such as fixed R&D and capital investments). However, the introduction of these costs would still yield the presence of an excessive/insufficient number of firms under the same parameter conditions as in our model.

Following our numerical example in previous sections of the paper, Fig. 4 considers costs \(c=1/4\) and \(z=1/2\). Other parameter values yield similar qualitative results and can be provided by the authors upon request.

For this numerical example, outcome \((G_{1},NG_{2})\) arises under condition \( \min \{K_{a},K_{c}\}<K<\max \{K_{b},K_{c}\}\), which in this case implies that \(K\ \)satisfies \(K_{c}<K<K_{b}\); as depicted in the right-hand side of Fig. 4.

While the shaded areas describe optimality in terms of the number of firms developing the green good, the externality that both types of products generate is still not addressed by any policy tool (such as taxes or quotas), ultimately implying that these areas only identify second-best optima. (Nevertheless, and for compactness, we refer to shaded regions as optima.)

In particular, cutoffs \(K_{a}\) and \(K_{c}\) lie on the negative quadrant. Thus, for all \(K>K_{b}\) the outcome \((NG_{1},NG_{2})\) is socially optimal, which only coincides with the equilibrium outcome in region III (in region I and II one or both firms develop the green good). For all \(K\le K_{b}\), outcome \((G_{1},NG_{2})\) is optimal, thus coinciding with the equilibrium outcome in region II alone.

For the parameter values in Table 1, where \(d=1/2\), the welfare benefit from consumer and producer surplus is exactly offset by the environmental damage from the brown product, thus yielding a zero welfare level. Other numerical simulations with \(d<1/2\) yield positive welfare levels, and can be provided by the authors upon request.

Note that this is not necessarily the case when goods are relatively differentiated.

Graphically, when \(\lambda =0.1\) (in the left-hand side of Fig. 9), moving from point \(A\) to either \(B\) or \(C\) entails a welfare reduction. Similarly, when \(\lambda =0.3\) (in the right-hand side of Fig. 9), moving from \(D\) to either \(E\), \(F\) or \(G\) yields a welfare loss.