Market and non-market policies for renewable energy diffusion: a unifying framework and empirical evidence from China’s wind power sector Authors: Yang LIU and Taoyuan WEI Published in Energy Journal Annex 1. Proof of Eq. 4. Since lim Qt−h → Qt , Eq. 3 can be expressed by β→0 π π‘ π·πππ‘ = γ · ππ‘ (1 − Qmax ) (A1) as β → 0. Notice that Eq. 2 can be rewritten as ππ‘ − ππ‘−β = ππ‘−β π·πππ‘−β ππ‘−β which is equivalent to πππ‘ 1 πππππ‘ π·πππ‘ = = ππ‘ ππ‘ ππ‘ ππ‘ when β → 0. By inserting Eq. A1, the above equation becomes πππππ‘ ππ‘ = π·πππ‘ ππ‘ π π‘ = γ · (1 − Qmax )= π[γ·π‘] ππ‘ π‘ γ − Qmax π(∫0 ππ£ ππ£) ππ‘ = π[γ·π‘] ππ‘ γ − Qmax π(πππ‘ ) ππ‘ , π‘ πππ‘ = ∫0 ππ£ ππ£ is the cumulative capacity at time t1. Hence, ππ‘ = π ππ π‘ ) γ·(π‘− max Q ππ¦ ππ π π’ππππ π0 = 1 and π‘ ≥ 0. (A2) By inserting Eq. A1 into Eq. 1 and rearranging terms, we have γ Qmax ππ‘ 2 + (1 − γ)ππ‘ − 1 1 1 1 +( − πππ₯ )·π −π·ππππ‘ ππππ₯ ππ‘ π =0 (A3) A lagged cumulative stock πππ‘−1 at time t is used in our empirical estimation. However, due to a continuous time framework, we cannot state t-1 in our theoretical model. 1 1 where If π πππ₯ is large, then γ 1 Qmax < ππππ₯ ≅ 0. At the earlier stage, we would expect ππ‘ is trivial compared to Qmax 2. Hence, the squared term of ππ‘ can be ignored and the above Eq. A3 can be simplified as (1 − γ)ππ‘ − ππ‘ · π π·ππππ‘ ≅ 0 (A4) By inserting Eq. A2 to Eq. A4, we obtain Eq. 4. Annex 2. Calculation of customer benefits for numerical simulation in Section 6 The customer benefits are calculated by πΌππ£ππ π‘ (ππ ππππππ¦) πΆπ΅π‘ = [πΆπ‘ − πΆπ‘πΌππ£ππ π‘ ] + ∑20 π=1 ππππππ‘πππ (ππ ππππππ¦) πΆπ‘ (1+π)π ππππππ‘πππ −πΆπ‘ (A5) where π is the average lifetime period of the wind farm; πΆπ‘πΌππ£ππ π‘ is capital costs and πΆπ‘ππππππ‘πππ is the O&M costs. With the common learning curve, we specify the investment and O&M costs as following: QS πΆπ‘πΌππ£ππ π‘ = πΆ0πΌππ£ππ π‘ · ( QS t )−π½ 0 πΆπ‘ππππππ‘πππ = πΆπ‘πΌππ£ππ π‘ · α where πΆ0πΌππ£ππ π‘ and QS0 are, respectively, capital costs and cumulated installed capacity at the starting point; π½ is the learning-by-doing coefficient; α is a parameter determining average annual O&M costs as a percentage of capital costs of a wind farm. In the case of China’s wind power 2004-2011, the maximum Q t is only 8.5 MW (Table 1) while the estimated Qmax is 0.45/0.00009=5000 MW for a province based on the estimated coefficients of π‘ and QSt in Model A (Table 2). Hence, the squared term in Eq. A3 is estimated to be only 0.45/5000*8.5 2=0.0065 even for the maximum Q t . 2 2 Annex 3: Evolution of different components of social welfare Fig. A1. Environmental benefits, customer benefits and subsidy costs (γ =0.38, Unit=billion RMB) 500 450 400 350 300 250 200 150 100 50 0 -50 201020112012201320142015201620172018201920202021202220232024202520262027202820292030 -100 Environmental benefits Customer benefits Subsidy cost Fig. A2. Environmental benefits, customer benefits and subsidy costs (γ =0.05, Unit=billion RMB) 300 250 200 150 100 50 0 -50 -100 Environmental benefits Customer benefits 3 Subsidy cost