Reference pricing (RP) theories predict different outcomes when reference prices are fixed (exogenous) versus being a function of market prices (MPs) (endogenous). Exogenous RP results in MPs at both high-price and low-price firms converging towards the reference price from above and below, respectively. Endogenous RP results in MPs at both high-price and low-price firms decreasing, with low-price firms acting strategically to decrease the reference price in order to gain market share. We extend these models to a hospital context focusing on insurer and consumer payments. Under exogenous RP, insurer and consumer payments to low-price hospitals increase, and insurer payments to high-price hospitals decrease, but predictions regarding consumer payments are ambiguous for high-price hospitals. Under endogenous RP, insurer payments to high-price and low-price hospitals decrease, and consumer payments to low-price hospitals decrease, but predictions regarding consumer payments are ambiguous for high-price hospitals. We test these predictions with difference-in-differences specifications using 2008-2013 data on patients undergoing joint replacement. For 2years following RP implementation, insurer payments to high-price and low-price hospitals moved downward, consistent with endogenous RP. However, when the reference price was not reset to account for changes in MPs, insurer payments to low-price hospitals reverted to pre-implementation levels, consistent with exogenous RP. Copyright (C) 2015 John Wiley & Sons, Ltd.