Hidden Value: How Consumer Learning Boosts Output

BY LEONARD NAKAMURA

phones. Ipads. Wikipedia. Google Maps. Yelp. TripAdvisor. product exists and then what its char- New digital devices, applications, and services offer advice acteristics and performance are like. and information at every turn. The technology around This information acquisition in turn I us changes fast, so we are continually learning how best lowers the risk associated with any to use it. This increased pace of learning enhances the given purchase and, on average, will satisfaction we gain from what we buy and increases its value to us over raise the amount of pleasure or use we time, even though it may cost the same — or less. However, this effect get from it. of consumer learning on value makes inflation and output growth more Consider all the information avail- difficult to measure. As a result, current statistics may be undervaluing able to help us decide to see a movie. household purchasing power as well as how much our economy We can look at trailers in the theater or online; we can read reviews and produces, leading us to believe that our living standards are declining compare the number of stars the movie when they are not. gets from critics or fellow moviegoers; and we can ask our friends. Similarly, when deciding on a restaurant, we can This disconnect has implications Then we will turn to theories of con- consult online sources like Yelp, Zagat, for policy. Economists are more famil- sumer preferences and behavior that or Chowhound; we can examine the iar with how learning makes us better take learning into account. They may menu and prices; we can read a review workers by increasing our productiv- point us toward more accurate ways to in the local paper; and we can listen ity, typically reflected economywide estimate inflation and output growth to our friends’ suggestions. All this in higher inflation-adjusted wages than measuring prices directly. information-gathering raises the prob- and output per capita. However, how ability that we will enjoy the movie or learning makes us better consumers MORE BENEFIT PER restaurant more than if we had chosen is less likely to be captured by official DOLLAR SPENT blindly. When we take the time to measures of consumption and out- In this era of rapid innovation find out more information, we are able put. To the extent that these statistics and creativity, consuming so many to select products most suited to our might be imprecise, economists are new products typically involves learn- tastes and will generally experience liable to be led astray in assessing the ing both before and after we purchase higher satisfaction per dollar spent, economy’s successes and failures, and them for the first time. Acquiring in- given a fixed menu of choices, than we policymakers may be misled in decid- formation about a product we haven’t otherwise would. Raising our satisfac- ing which actions to adopt. bought before is so automatic that we tion per dollar may also make us more But how can one measure the im- may hardly notice it as an economic willing to buy more products within pact of consumer learning on the well- phenomenon. Indeed, if the product that category. being of households? First, we need to is novel, we must acquire at least some A second layer of benefits occurs explore just how learning affects value. information: First we find out that the through use: Using the features on my e-mail or word processing program becomes second nature as, one by one, I try out new tasks. This form of Leonard Nakamura is a vice president and economist at the Federal learning-by-doing raises the product’s Reserve Bank of Philadelphia. The views expressed in this article are not necessarily those of the Federal Reserve. This article and value in later uses; once I know that other Philadelphia Fed reports and research are available at www. a feature exists and how to use it, I philadelphiafed.org/research-and-data/publications. can more quickly find it and use it. As I learn to use my smartphone by www.philadelphiafed.org Business Review Q3 2014 9 making a call or finding a destina- mation allows the smart consumer to has gone up, then this is not the right tion or taking a picture or watching a choose movies, TV shows, restaurants, measure of our inflation rate, since the video clip, using it becomes faster and and a myriad of consumer products quality of the service has risen and we more successful.1 Moreover, with cheap and services that are more to our lik- get more for the price. memory and computing power, we can ing. The cost of the better information Similarly, our cable TV bills (as customize the devices and applications that helps us make these better choices measured in the U.S. CPI index of “ca- to our needs. Using an application can has fallen, allowing us to derive greater ble and satellite TV and radio”) have also result in a valuable history to tap satisfaction from what we buy. Thus, risen at an average annual rate of just later: The letters I have written and our knowledge of the Internet enhanc- over 2 percent over the past five years. the PowerPoint slides I have produced es the value of — and spurs the devel- Does this rate fully reflect the greater in the past may have pieces that I can opment of — new ways to reach it. value we derive from cable service? insert into new e-mails and presenta- Yet, so much of the content on the When we first use cable TV, we may tions. In many cases, the application has the ability to learn our habits and guide us to better choices, sometimes Does this improvement in our welfare show up using the preferences of other users in measures of real consumption and growth? who make choices similar to ours. For example, Netflix looks at our past Typically not. movie choices to suggest new ones. What is economically significant Internet — videos, TV shows, music, know only a few channels. Over time, about this form of learning is that and social media — is available at no as we channel-surf and learn more the product is the same, but we value extra cost. So, as we learn about the about the content shown on different it more. Yet, standard measures of Internet, we use our connection to channels, we may become attached to economic output miss this increase it more intensively, but we don’t pay three or four channels we didn’t know in value because the product appears more. The Internet connection itself about before. As a result, access to unchanged. As a result, statistics is unchanged; what is changed is the cable TV becomes more valuable to us. measuring overall consumption may content and interactions it gives us But how can we measure that value? be too low.2 access to. Because if the satisfaction For example, let’s consider how we we gain from the Internet connec- MEASURING THE VALUE value an Internet connection. Entre- tion is greater, we would be willing to OF INFORMATION preneurs keep developing search en- pay more for it. But if the market for Consider a traveler planning to gines, aggregators, instructional sites, Internet connections is competitive, go to a foreign city for the first time. and various applications that make we don’t have to: Competition prevents Initially, the traveler sees that hotels A our use of the Internet more efficient. providers from charging more as Inter- and B are equally priced and have sim- Plus, smartphones and tablets make it net offerings expand, so we get more ilar luxury levels as measured by that easier to connect whenever we want value for the same amount of money. country’s rating scheme. But the In- and wherever we are. All of this infor- But does this improvement in our ternet allows the traveler to see reviews welfare show up in measures of real from other travelers, detailed maps of consumption and growth? Typically the hotels’ locations, and lists of the not. The monthly fee we pay to the In- hotels’ amenities. Let’s say that the 1 Although this article does not explore the notion, it must be admitted that there is a ternet service provider this year is buy- more knowledgeable concierge at hotel countervailing truth: Our existing knowledge ing more for us than the monthly fee B is worth $10 a day to the traveler. may become outmoded at a faster rate as new technologies race at us. This depreciation of we paid five years ago. If the fee has Learning about the concierge over the our knowledge is a cost of rapid technological gone up, we measure this as pure infla- Internet makes the traveler better off progress but is also something we have difficulty tion: The price of “Internet services by $5: In the absence of this informa- measuring. and electronic information providers” tion, the traveler would have chosen 2 Another interesting implication of consumer in the U.S. Bureau of Labor Statistics’ randomly between the two hotels and learning is that it may be one reason that so- called early adopters are willing to pay a higher consumer price index (U.S. CPI) has would have gotten the good concierge initial price for the latest technology. Even gone up at an annual rate of 1 percent. half the time, for an expected value of though they realize the price will drop later, they know they will become better off as they But if the satisfaction we have gained $5. But with the information obtained learn more about the product. as we use the Internet more intensively from the Internet, the traveler gets

10 Q3 2014 Business Review www.philadelphiafed.org the good concierge all the time, for an or she would be just as happy with one vices would cost in the second year. expected value of $10. With better bundle as another. It is this assumption This provides us with a measure of the knowledge, the traveler gets more satis- — that we can find bundles of products rate of inflation the consumer faces. faction from the same set of choices at across which consumers are indifferent Alternatively, we can measure the set the same price. Here we can quantify — that economists rely on to estimate of goods and services the consumer the improvement as $5. The traveler inflation and economic growth. We actually bought in the second year and knows how much to value the con- will then discuss how behavior is dif- ask how much that set would have cost cierge and would have been willing to ferent in situations in which learning in the first year. This second measure pay $10 more to stay at that hotel than is occurring and how these changes in of inflation is typically lower than the at the other. behavior influence pricing and welfare. first one.4 We can use either measure, But measuring this value may or we can average the two. require new methods. Statistical GENERALIZED UTILITY If we believe that consumers have agencies charged with measuring FUNCTION THEORY stable preferences over these prod- prices usually simply ask the hotels In a classic 1977 article, “De Gus- ucts — that is, more or less unchang- what prices they charge. Instead, an tibus Non Est Disputandum,”3 George ing utility functions — then we can agency might have to survey consum- Stigler and Gary Becker argue that say that if consumers’ incomes in the ers to elicit these evaluations. Alter- human tastes are fundamentally the first year rise at the rate of inflation, natively, Internet-savvy hotel opera- same; they “neither change capri- consumers could afford to buy approxi- tors or tourist organizations could do ciously nor differ importantly between mately the same goods and services experiments to elicit the extent to people.” Where it appears that tastes they had bought the year before and which customers are willing to pay vary, Stigler and Becker widen the are just as well off. We then can more for superior information. notion of consumer preferences from say that their real incomes haven’t The effect of learning on value specific goods and services to broad, changed. If their incomes are 2 per- isn’t limited to technology. For in- unchanging categories that they call cent higher than the rate of inflation, stance, learning to play an instrument commodity objects of choice. These we say that their real incomes have often deepens our understanding and stable preferences have goods and ser- risen by 2 percent, because they can enjoyment of music. The information vices as inputs, but also the consumer’s buy 2 percent more than they could we gain isn’t only steering us to the time and human capital such as educa- the year before. But if consumers’ util- music we prefer; it also deepens our tion and the acquisition of informa- ity functions change over time, this appreciation of the music. We make tion. Thus, individuals can actively claim might become dubious: If last a human capital investment that im- shape the satisfaction they derive from year I liked fish and bought a lot of it, proves our ability to consume, similar specific goods and services by obtain- and this year I don’t like it as much to a long-term investment in a home or ing knowledge. But Stigler and Becker but still buy a lot because it is cheap, an education that makes us better able point out that this broader way of look- then I may be worse off, though I am to earn a living. Here we might wish ing at preferences changes the nature buying the same amount. To be sure, to quantify the investment in infor- of income and prices. our preferences may fluctuate; I may mation that consumers make in order Stable preferences are key to prefer fish one year, meat another. But to quantify the value of the informa- measuring inflation. Ordinarily, if these back-and-forth changes may not tion, in the same way that we might we can identify bundles of consumer matter to our overall measures if these measure a consumer’s investment in a goods and services about which a fluctuations cancel out — for every home or a car. consumer is indifferent in two succes- individual who likes fish less, another To analyze consumption when sive years, this starts us on the way to likes it more. What Stigler and Becker learning is occurring, let’s first explore estimating inflation and output growth were concerned with were systematic some underlying theory regarding esti- between the two years. We first look changes in taste. mating changes in prices and output. at what the consumer actually bought This theory will allow us to construct in the first year and then ask how a stable “utility function,” a method of much that exact set of goods and ser- 4 The bundle bought in the second year is typically cheaper because goods and services representing consumer preferences that increase in price at different rates, and consum- permits us to assume that there are ers tend to buy less of the more expensive goods. bundles of products and services across So the second year’s purchases will typically 3 Translatable as “There’s No Arguing About have fewer of the goods whose prices rose more which a consumer is indifferent: He Taste.” rapidly. www.philadelphiafed.org Business Review Q3 2014 11 The generalized utility func- enjoy it less. Another implication of pharmaceuticals under patent typically tion is stable. To demonstrate how this perspective is that when we are rise in price faster than inflation. Even underlying preferences may be seen to young, we may not like a certain type absent monopoly, learning is one of the be stable, Stigler and Becker cite what of music very much initially, but we main reasons why customers may find appears to be an example of a changing may realize that we will gain human it difficult to switch from one supplier utility function: addiction — the phe- capital that will make the early invest- to another.6 nomenon that “smoking of cigarettes ment worthwhile in retrospect. … or close contact with some person Note that a given act of consump- MEASURING INFLATION over an appreciable period of time tion — for example, listening to or AND OUTPUT often increases the desire (craving) for playing music — may have both an There are two ways in which we these goods or persons.” But if we re- aspect of direct consumption (our can be better off economically: We can formulate the specific product cigarettes current enjoyment) and an aspect of have more products and services, or we into the broader commodity smoking, investment (how our current con- can make better use of what we already or close contact into the commodity lov- sumption affects our future enjoy- have. It is easier, however, to measure ing, perhaps we can understand them ment). Both aspects increase our quantity than quality. To think this as stable human behaviors.5 current willingness to pay for the item. through, consider how we currently Citing Alfred Marshall’s example This makes for interesting dynam- measure output and inflation. of music — “The more good music ics over time. As we age, the period Suppose I spent $20,000 on a man hears, the stronger is his taste over which our investment will pay consumer goods and services in 2013 for it likely to become.” — Stigler and off shortens, but our enjoyment rises and $21,000 in 2014. Is my well-being Becker argue that an individual can because of past learning. Eventually, higher in 2014 than it was in 2013? accumulate “consumption capital” in though, our rate of learning and the The test that economists normally use music, so that, for instance, buying rate of increase in enjoyment slow is to ask whether I could have bought tickets to a concert at one point in time increases the satisfaction derived from further consumption of music later. Thus, just as workers can invest Just as workers can invest in education to in education to enhance their produc- enhance their productivity at making objects or tivity at making objects or providing providing services, so can consumers invest services, so can consumers invest in education to enhance their enjoyment in education to enhance their enjoyment of of certain goods and services. This in- certain goods and services. creasing satisfaction can be understood as “rational addiction,” in that consum- ers can understand and predict ratio- down, so we are less willing to pay the same goods and services in 2014 nally how their consumption in one because the investment value is falling, as I bought in 2013. If so, I must be at period may affect their consumption in even though our direct enjoyment is least as well off, because I could have future periods. Thus, I can decide not still increasing. bought the same goods but didn’t. to consume a drug that I know I will As we become more willing to Therefore, I must have preferred the enjoy this period but that will induce a pay for something, do we have to pay goods I did buy to the goods I didn’t, craving in future periods, when I will a higher price? A drug dealer may since I can freely choose what I buy. offer the first dose of a drug for free, So I strictly prefer what I consumed in hopes the customer becomes ad- this year to what I consumed last year. 5 In another example they explore, Stigler and dicted. This depends on there being However, as we have seen, when Becker view advertising as a means of providing information to consumers that improves their some likelihood that the person offered consumers learn about a product, it perceived benefit from the product being ad- the free drug will remain a customer vertised. In this case, the maker of the product of the dealer, so that the addiction provides information that changes the value of the commodity consumed. They also discuss can be exploited. If the producer has 6 As we use products and services, our learning fads and fashions and the role of culture and a monopoly on the good whose value may result in what are known as increased traditions in the formation of tastes. See my to us has increased, then the price switching costs. See Paul Klemperer (1995), Business Review article on advertising for further Carl Shapiro and Hal Varian (1999), and Luis discussion. may rise over time. This may be why Cabral (2014), among others.

12 Q3 2014 Business Review www.philadelphiafed.org can provide more satisfaction than Willingness to pay. Alternatively, the rate of inflation and the increase it did initially. In this case, we may we can attempt to directly measure in income necessary to keep our wel- want to consider my consumption as how the consumer’s willingness to fare constant. We may think that liv- having increased, even though what I pay has changed. For example, if the ing standards are falling when they are, consumed did not change physically. price of a good rises and the consumer in fact, rising. After all, when we dis- But if the good or service in question consumes as much of that good as she cuss how we might raise productivity is unchanged, how do we measure the did previously, or if the price remains or consumer welfare, we typically rely increased satisfaction it offers, that is, the same and the consumer consumes on our existing measures of output and its increased utility? There are at least more of the good, then we may be able inflation. But to the extent that we two routes that we might take. to measure an increase in the consum- think we might be getting this measure Consumer investment in er’s willingness to pay. wrong, we might decide to temper or consumption. One view is that in Consider pharmaceuticals. Sup- slant our objectives. For example, how learning about, say, music, consumers pose the efficacy of a drug improves we think of price stability is tempered are investing by directly raising the sat- over time as doctors and patients by beliefs that our inflation measures isfaction they receive from music. In share information about its effects are likely subject to a measurement principle, an investment in consump- and as treatment regimens are fine- bias, and we have a rough idea of the tion is no different from an investment tuned accordingly. We may be able to size of that measurement bias. As a in durable consumer goods, such as directly measure the drug’s increased consequence, a small but positive in- cars and refrigerators, or in real estate, value to both doctor and patient as flation rate may be viewed as achieving such as a single-family home. Any in- a result of this social learning. A price stability. vestment is expected to return value to similar case can be made for medical But it would clearly be desirable the investor — either in cash or well- procedures. An interesting possibil- if economic statistics measured output being — over an extended period. ity is that a given intervention — for and inflation more accurately. The If we are learning about a tech- example, use of a checklist in anes- report of the Commission on the Mea- nology that we expect will be around thesiology or surgery — may result surement of Economic Performance for a long time, then our learning may in a widespread improvement in the and Social Progress seeks to move na- be valuable for a long time. Just as an quality of medical care.7 Again, as tional statistical measures closer to an investment in understanding music is the intervention becomes widely ad- ideal measure of progress in national likely to bear fruit over an entire life- opted, we may be able to measure the well-being. The commission’s report time, so may an investment in touch- joint value of this social learning as points out that policymakers and typing, which enhances the speed and the quality of a variety of treatments others use these statistics to measure accuracy with which we can write e- (different surgeries, say) improves. economic success. To the extent that mails and Internet posts. Even though current statistics are biased, policy- the specific items we purchase — PCs, CONCLUSION makers are liable to be led astray. tablets, smartphones — may last only Does measuring the benefits — Thus, it would be valuable to consider a few years, touch-typing is valuable and the costs — of consumer learning how best to measure the impact of in using all of those products and may matter, particularly if they are difficult education, learning, and information enhance our ability to communicate to measure accurately? Even if econo- on the well-being of households and to over many years. mists cannot put numbers on them, it incorporate these measurements into So to measure the increased satis- is important to understand the limits our statistics. As new technology and faction gained from such a consump- of what can be measured. If we cannot learning make measuring inflation tion investment, we want to measure measure the improvement in our well- and output growth more difficult, we both the money and the time invested. being from learning about products, may not be able to rely on direct price Then we want to estimate the rate of then we underestimate our progress as measures; rather we may have to use return on those investments. Because consumers, and we overestimate both surveys or econometric methods to we need to know over what period of estimate inflation and growth. BR time the investment will create returns 7 Atul Gawande, a surgeon and journalist, has and how much consumers value those written about this in his book The Checklist returns, we have to survey consumers. Manifesto. 8 See Stiglitz, Sen, and Fitoussi.

www.philadelphiafed.org Business Review Q3 2014 13 REFERENCES

Cabral, Luis. “Dynamic Pricing in Custom- Marshall, Alfred. Principles of Economics. Shapiro, Carl, and Hal R. Varian. Infor- er Markets with Switching Costs,” working London: Macmillan, 1890. mation Rules. Boston: Harvard Business paper (April 2014), http://luiscabral.org// School Press, 1999. economics/workingpapers/scostsApril2014. Nakamura, Leonard. “Intangible Invest- pdf. ment and National Income Account- Stigler, George, and Gary S. Becker. “De ing: Measuring a Scientific Revolution,” Gustibus Non Est Disputandum,” Ameri- Gawande, Atul. The Checklist Manifesto. Review of Income and Wealth, S1 (2010), can Economic Review, 67 (1977), pp. 76-90. New York: Metropolitan Books, 2009. pp. 135-155. Stiglitz, Joseph, Amartya Sen, and Jean- Klemperer, Paul. “Competition When Nakamura, Leonard. “Underestimating Paul Fitoussi. Report of the Commission on Consumers Have Switching Costs: An Advertising: Innovation and Unpriced the Measurement of Economic Performance Overview with Applications to Indus- Entertainment,” Federal Reserve Bank and Social Progress (2009). trial Organization, Macroeconomics, and of Philadelphia Business Review (Fourth International Trade,” Review of Economic Quarter 2005). Studies, 62 (1995), pp. 515-39.

14 Q3 2014 Business Review www.philadelphiafed.org