We provide a solution that may offer closure to the question on how to measure the empirical relation between stock market values and accounting numbers. The models that dominate studies of the relevance of accounting numbers produce coefficient estimates that are hard to interpret and exhibit great volatility. We present a theory that demonstrates a multiplicative power law describes the long-run relation between market values and accounting values. Consequently, the correct forms for estimating the market-accounting relations are log-linear. Crosssection models based on this theory produce elasticities that are valid and accurately reflect the value relevance of accounting variables. We estimate these elasticities in the cross-section for the years 1971-2016. We compare our multiplicative model, log-linear, estimates to response parameters of traditional, additive-linear models that relate market and accounting values. Our results demonstrate the superiority of using elasticities to measure the empirical relation between market values and accounting numbers.