The fact that income and substitution effects move in opposite directions in the case of inferior goods raises a tantalizing possibility: What if the income effect were the stronger of the two? Could demand curves be upward sloping?
The answer, from a theoretical point of view, is yes. If the income effect in Figure 7.5 "Substitution and Income Effects for Inferior Goods" were larger than the substitution effect, the decrease in price would reduce the quantity demanded below q1. The result would be a reduction in quantity demanded in response to a reduction in price. The demand curve would be upward sloping!
The suggestion that a good could have an upward-sloping demand curve is generally attributed to Robert Giffen, a British journalist who wrote widely on economic matters late in the nineteenth century. Such goods are thus called Giffen goods. To qualify as a Giffen good, a good must be inferior and must have an income effect strong enough to overcome the substitution effect. The example often cited of a possible Giffen good is the potato during the Irish famine of 1845–1849. Empirical analysis by economists using available data, however, has refuted the notion of the upward-sloping demand curve for potatoes at that time. The most convincing parts of the refutation were to point out that (a) given the famine, there were not more potatoes available for purchase then and (b) the price of potatoes may not have even increased during the period!
A recent study by Robert Jensen and Nolan Miller, though, suggests the possible discovery of at least one Giffen good. They began their search by thinking about the type of good that would be likely to exhibit Giffen behavior and argued that, like potatoes for the poor Irish, it would be a main dietary staple of a poor population. In such a situation, purchases of the item are such a large percentage of the diet of the poor that when the item's price rises, the implicit income of the poor falls drastically. In order to subsist, the poor reduce consumption of other goods so they can buy more of the staple. In so doing, they are able to reach a caloric intake that is higher than what can be achieved by buying more of other preferred foods that unfortunately supply fewer calories.
Their empirical work shows that in Hunan province in southern China rice is a Giffen good for poor consumers. Rice provides calories at a relatively low cost and dominates the diet, while meat is considered the tastier but higher cost-per-calorie food. In order to look at individual household decision making, they conducted a field experiment in which randomly selected poor households were given vouchers, redeemable with local merchants, for price reductions of varying sizes on the staple good. Households and merchants were given explicit instructions that selling the vouchers for cash or reselling the staple good would result in dismissal from the program and audits of the program seemed to confirm that participants were conforming to the ground rules. Overall about 1,300 households participated. Households also completed a detailed questionnaire reporting what they ate and drank, as well as other characteristics of the family on income, employment, other expenditures, and the like. They then divided the households into two categories: 1) those who were so poor that, prior to the experiment, almost all of their calories were from the staple good (Households in this category would not be expected to show Giffen behavior because their extreme poverty gives them no choice but to consume less of the staple when its price rises). and 2) those who were somewhat less poor in the sense that, prior to the experiment, they got at least 20% of their calories from sources other than the staple good. Households in this "poor-but-not-too-poor" group exhibited Giffen behavior. In particular, they estimated that a 1% increase in the price of rice leads to a 0.45% increase in rice consumption.
A similar experiment by the authors on wheat consumption in Gansu province in northern China showed less evidence of its being a Giffen good, probably because there are more substitutes available for the specific form of wheat - wheat flour used to make wheat-based foods in the home - that was the subject of the experiment. In Gansu, people also consume wheat noodles at restaurants or road-side stands or buy wheat-based products from stores in prepared forms. A study by David McKenzie tested whether tortillas were a Giffen good for poor Mexicans. He found that they were an inferior good but not a Giffen good and similarly speculated that the availability of substitutes was the likely reason.
Jensen and Miller argue that despite the fact that their research is the first to uncover a real example of a Giffen good, other examples are likely waiting to be discovered in areas of the world where the population is poor but not-too-poor and where there are few substitutes for the staple good.