LvW reminded me, that my previous comments ignored reasonable circuit operation, because I restricted myself to answering the raised question only.
Applying the criteria explained by LvW to the said circuit, you realize, that's a rather bad design. Voltage divider and capacitively bypassed emitter resistor suggests case (b), current controlled voltage feedback. But unfortunately the emitter resistor value is much too low a achieve a noticeable feedback (a few 10 mV voltage drop across the resistor). In return, the voltage divider has a too high impedance of about 600k to source a constant reference voltage. Effectively, the circuit stabilization effect isn't anything better than a single resistor between Vcc and base, omitting the emitter resistor. This circuit works only, if the resistor value is matched to the transistor current gain. This circuit, base current injection without feedback is the second worse biasing method, the worst would be voltage injection without feedback.
To change the circuit operation to more stable bias, you would either apply said method (a) by connecting the feedback resistor to the collector instead to a voltage divider, or method (b) by increasing R18 and omitting R16, keeping the lower divider resistance level.