Yes, you can use PWM to control a pump (although indirectly), but it's not as straightforward as you think.

skimask has got close. Use the pump to maintain pressure in a hydrosphere (pressure vessel) at a higher pressure than you need (say 180 or even 200 psi). The pressure vessel feeds a Pressure Reducing Valve (PRV) which is set to your constant pressure. So if you need 150psi, you set the PRV to deliver that, but the high-side (input) to the PRV must always be greater than the set pressure. The pressure vessel is there to take-up the slack (like a Capacitor), so if you need a small amount of liquid (say to fill a kettle) the pressure vessel delivers it rather than kicking-in a pump to deliver half a glass of water. The secret is to install the biggest pressure vessel that space and money allows. The smaller the vessel, the more your pump is going to run. You then also set the largest differential (the difference between the On and Off pressure settings running the pump). So in the above example, pump-ON at 155psi and OFF at 200psi is great.

If you're feeding water to the top of a building several stories high, then 200psi is quite normal. Usually, you need about 15-20psi per floor. If your roof tank is 10 floors up, then yes, 200psi.

As another (expensive) option, you can get PRV's that can be modulated by a PWM signal. That's not really for a constant-pressure application, but we use such devices in the Laboratory to give a broad range of pressures (all adjustable by a PIC) for testing and calibration of pressure sensors and transducers.