With the rise in " from micro to macro” paradigm recently, the information of aggregate earnings gets much attention and its predictability for economic growth has been confirmed. As Ball and Sadka（2015）points out, however, the predictability of aggregate earnings is not the only accumulation of firm-level earnings, researchers should focus on the effect paths of forecasts of macro-economic growth by accounting information. Based on the intrinsic logic of national economic accounting system and accounting conservatism principle, this paper decomposes earnings into assets impairment losses（AIL）and adjusted accounting earnings（AAE）, and investigates the paths of forecasts of future GDP growth by accounting information. The theoretical logic for our design is following. The accounting earnings are the matching result of income and costs using the accrual basis, which reflect the events and transactions in the past, and its measurement attribute is mainly historical costs. As the component of GDP, operating surpluses are highly compatible with accounting earnings in concept and calculation. We thus propose that the predictability of aggregate earnings largely drives from the mathematical correlation of accounting earnings and operating surpluses. In other words, there might be a path function as " current AAE-future operating surpluses-future GDP growth”, which is defined as " earnings conduction”. However, the predictability of AIL follows a different logic from AEE. First, the AIL embodies the conservatism principle, which requires that the expected future losses should be recognized in advance. Therefore, the AIL item in annual reports indicates the managerial estimation about the business risks and negative macro shocks, and manifests the consequences of fair value measurement. Second, the AAE is highly serially correlated, while the recognition of AIL is accidental and is excluded from the national economic accounting system. Third, as the AIL is the result from managerial judgment and estimation on risks, it contains much private information about macro risks. The AIL might be manipulated for earnings management purpose; nevertheless the aggregation progress could offset the noises and thus reflects systematical estimation of further macro-economic risks. Therefore, we define the predictability of the AIL driving from the systematical estimation of future macro-economic risks as " risk perception”. Based on the aggregately quarter data of Shanghai and Shenzhen A-share companies from 2003 to 2016, this paper confirms that both AIL and AAE could predict future GDP growth, but both predictability follows different paths. By adopting the mediating effect model, the macro-economic risk has a significantly mediating effect in the prediction process of AIL, but not significant in the prediction process of AAE, confirming the " risk perception” path. Meanwhile, the reverse regression tests show that the predictability of AAE mainly comes from its mathematic correlation with operating surpluses, and the correlation of AIL and GDP growth is not affected by operating surpluses, which is consistent with the " earnings conduction” path. Furthermore, we find that the negative relation between AIL and future GDP growth is stronger during economic downturn period than during economic upswing period, while the prediction function of AAE has no difference during economic downturn or upswing periods. The potential contributions of this paper are as follows. First, through integrating the conservatism principle with the national economic accounting system, we firstly propose and demonstrate the effect paths of forecasts of macro-economic growth by accounting information: " earnings conduction” and " risk perception”. Second, a stream of literature investigates the micro-level consequence of conservatism principle and also studies its relation to AIL, but little examines its macro-level consequences from the macro-perspective except Crawley（2015）that finds accounting conservatism could impact monetary policy. This paper provides further evidence for the macro-economic consequences of accounting conservatism.