# Of key importance is the interpretation of

We posit that Leflunomide the SMP is the marginal cost based on the competitive bidding requirements of the SEM (lnep2_smp in Table 1).equation(3)?lnC?lnYD=lnSMP

We present estimation results from three models in Table 2 below. The estimation methods are Prais-Winsten regressions corrected for first-order serial correlation of the errors (AR1).10 The rationale for presenting three models is mainly for specification sensitivity. Models M1 and M2 are the same estimates of equation (2) with different sets of seasonal dummies: M1 contains selected months for seasons, M2 and 12 (minus one) monthly dummies.11 M3, rather than including demand and wind generation separately, contains the difference between demand and wind generation, and its square (lndem_wind, lndem_wind2). M3 is segregation thus a specification sensitivity which restricts the coefficients on wind and demand to be equal, but allows for differences to be non-linear. This has some intuitive appeal, as wind generation is ‘must-run’ in the SEM, and in essence, dispatchable thermal generation is demand net of wind.