Note: I’ll be referring to the Gynoids/Fembots just as “machines”, out of personal preference (or lack thereof).
Features
Let's break down these machines by "features":
They can carry around objects, but they cannot clean nor cook.
This means, it lacks the fine dexterity to handle a hand tool.
They can easily be knocked down, even by a kid.
It either lacks strength or lacks balance. Definitively of little use as a guard.
Their surface texture has distinct "artificial" feel to it
So touch give it away, it won't pass as human if skin contact is required
and their "sexual" abilities are "acceptable" for most (single) men.
This is probably one of selling point for this product.
Intelligence-wise, they are no better than modern virtual assistants.
So, there exists technology that's better in the market. The designers didn't go for the most intelligent solution, either because it wasn't needed for the intended task or because it would have made it too expensive.
They can hold simple conversations, but anything more complicated will confuse them. Something like more advanced chatbot. So best they can do is to remember when their owner returns home, know his favorite food, and order that favorite food so it arrives right after he returns home.
This is the other selling point, they could be marketed as artificial companions.
While they could be useful to assist elderly people or even as baby watch... The fact that they have sexual abilities suggest that the market is single lonely male adults wealthy enough to pay.
Men with a bad job, that don't have time to date or don't earn enough to sustain a family could be the main market.
Employment
When introduced, as an expensive item, one of the first business models to use them is to rent them. The legitimate use would be to greet people in events, that would be expensive and sporadic.
Whoever considers that business will want to increase income by renting by the hour instead of days, adding an insurance fee, and a no-questions-asked policy... because we both know we don't want to ask what people actually do with these machines.
Note: below I use the word "programmed" loosely. Programming doesn't have to mean to write in an artificial language, it could be just setting configuration (like programming an alarm clock), some programming may mean to interface with a server elsewhere which is doing the actual work, or in some cases it could also mean to mod the device. Edit: perhaps it can be better to think about it as installing an "app".
The next stage is where people start to give more permanent tasks to the machines:
As receptionist/secretary: it can greet people to buildings, manage appointments, and it could even work as security camera and be programmed to trigger an alarm on unexpected behavior (e.g. unknown person entering without appointment, person entering with weapons, any activity after work hours, etc...).
As a teacher/lecturer: it could be programmed to repeat a speech, and even be able to answer a set of predefined questions.
As waiter: it could be programmed to deliver orders from the tables to the kitchen and to carry the dishes back to the table. They could recognize customers, be able to recite portions of the menu, and even answer simple questions about the dishes or do recommendations.
As guide: it could be used to make tours, to show people around a building, or even tours of a city if somebody or something else is driving (autonomous vehicle?)
Others: Cashier, Bank-tellers, Librarian, and other similar jobs could get a similar treatment.
Disruption
I'll assume that this product becomes popular in some areas where it is a good deal for business. All the impact will be local to those areas.
It will take a while since the introduction in the market to become popular... if at all. Some business will have to step up and take the risk to see how viable is to use them, and they will face the most opposition... if that goes ok, others will follow.
Addendum: Perhaps it doesn't become popular with the first bussiness that tries it, it just means that the product or its production will have to be improved so that it provides a better cost/benefit scenario. After about a generation (20 to 25 year) it probably will be good enough, and there will be new people willing to take the risk. Think Virtual Boy vs Oculus Rift.
Under the assumption that at the point where these are affordable and attractive for the single man, that person probably wasn't going to get a STD or cause an unwanted pregnancy... so, any effects on that is negligible. Although it may improve quality of life and even cut suicide rates in regions where those are problems.
On the economy, they are just another step toward replacing jobs with robots. We can model the effects using the Kübler-Ross model (a.k.a the stages of grief):
Denial: A) people don't see a threat in these products. They are just a novelty or a thing for weirdos that won't affect the economy overall. B) buying these things is seen as waste, and whoever does it is seen as a loser.
Anger: Some people start losing their jobs to be replaced by these machines, and people don't like it.
Bargaining: some people are willing to pay more to be attended by a real person, and some people are willing to earn less to keep their jobs. Labor unions negotiate ways to keep people employed, and laws are made to restrict the use of these machines. Soon, they need to be audited periodically by the government to check they are working correctly, adding extra cost to their use in business.
Depression: The jobs that were lost were lost, the times where you could get a human being at the counter were better. It is all mechanical now, no need for manners or respect. Just give up, machines win, humans are useless.
Acceptance: I'm unsure how this looks like, but people got to find another source of revenue. They have to accept that they have lost a portion of the job market to these machines.
Not really all jobs can be gone, it is easier to disrespect these machines than human being, and people need income, so there will more crime. Which means that security guards are in good demand.
The time frame for those stages isn't clear, although "bargaining" may last a few years just because bureaucracy and law making isn't exactly the faster industry.
Rural and industrial areas would have a lesser impact, because most if not all of the applications of these machines are in urban service and commercial areas. That would suggest that there has been no impact in traditional food industry.
Stores and markets may be cutting prices, after all people aren't earning as much and they are saving in payrolls. In some areas, it could also be a push for local food production (e.g. neighborhood farm).
So, I don't expect starvation or anything like that, people will settle to something else.
The value of the land will initially go down, as the affected areas are places where it could be hard to get jobs (people may want to move out to other areas where they can get job easily, so they may willing to sell cheaper). But it will eventually go back up.
Note: I did consider that there could be less new families in urban areas where these machine are popular, not because male prefer the machines as sex partners... but because income would have gone down due to lost jobs and human interaction would have gone down. Although I thing this effect is negligible too, because it won't affect all areas and because people will find solutions soon enough.