I am a little confused by this exercise. If you have household data for one period (let's say for the period 1997-98), how does the price enter as a variable? Does it vary regionally or for each household? Or do you have monthly data on each household so that in Table 1 each period signifies panel data with both months and households?

For 1999-2000, there may be multicollinearity since the price of potato is not significant although R square is not that high??

If a household substitutes potato for rice, we are talking about cutting back on rice and increasing potato intake but still consuming the two together. Still this is a new kind of low level 'potato famine' that you are uncovering.