Unexpected coefficients signs

Am doing a research on impact of stock liquidity on firm performance. I have a panel data of 81 firms over a period of 12 years from 2005 to 2016. I run Housman test to choose the proper estimation method for my model of which finally, i fall for fixed effect. I have Torbin Q as a proxy for my dependent variable (performance) and my independent variables include; Liquidity as main dependent variable, Firm age, Firm size,leverage, dividend pay ratio as control variables. Unfortunately,liquidity which is the main dependent variables turn out to be negative and insignificant contradicting my expectation.
Any hint out will be highly appreciated.