Amid widespread concern about the quality of the Chinese GDP data, power generation has become probably the single most cited indicator about the Chinese economy other than official statistics. For the first time, we provide an assessment on the efficacy of power generation as such an indicator, using both conceptual and empirical analyses. We show that (i) conceptually, it is somewhat misleading to use power generation as coincident indicator to infer the growth rate of GDP, due to the change of power intensity; (ii) empirically, power generation beats railway cargo and bank loan, the two other variables in the Li Keqiang index, as the best predictor in forecasting the Chinese GDP over the full sample; however, this superiority holds mainly during period of sharp change, i.e., the financial crisis period, but not when the economy is relatively stable, especially the current New Normal period. These two findings place important caveats on the common trust we have in power generation as an influential variable for measuring the Chinese economy.