Consumer Attitudes For Forecasting

Citation:

Richard T. Curtin (1984) ,"Consumer Attitudes For Forecasting", in NA - Advances in Consumer Research Volume 11, eds. Thomas C. Kinnear, Provo, UT : Association for Consumer Research, Pages: 714-717.

Advances in Consumer Research Volume 11, 1984      Pages 714-717

CONSUMER ATTITUDES FOR FORECASTING

Richard T. Curtin [Director, Surveys of Consumers, Survey Research Center, The University of Michigan]

Behavioral Economics

The Survey Research Center at The University of Michigan began a series of periodic consumer surveys more than 35 years ago, based on a theoretical view designated as behavioral or psychological economics. The underlying theory focuses on the human factor in economic affairs. Whether the economy falls into recession or moves toward recovery and growth depends on changes in business investment spending, consumer purchases, and government spending (aside from net exports and other factors). Keynes drew attention to the roles played by business expectations and the government sector, but it was Katona who first recognized and documented the role played by the consumer sector in determining the aggregate course of the economy (Katona, 1975). Behavioral economics views the consumer sector as having the same type of active and independent role attributed to the business community and to the government in shaping the macro economy. The consumer is a powerful actor, since the actions of consumers not only directly influence the course of the macro economy, but consumer spending and saving decisions can also act to reinforce or offset actions taken by business or government.

The power of the consumer in the macro economy resulted from three changes in the post World War II American economy. First, growth in consumer incomes as well as increases in liquid assets and other wealth holdings provided people with financial latitude in their reactions to changes in the economic environment. Financial latitude provided greater discretion in the range and timing of spending and saving decisions. Consumer reactions to changes in income could no longer be expected to result in immediate and offsetting changes in consumption. As a result, consumer attitudes and expectations became an import,ant ingredient for the analysis of fluctuations in consumer spending and saving.

The second change in the economic environment centered on the growth in consumer demand for housing, vehicles, and household durable goods. Consumer purchases of these goods account for an important share of consumer spending and overall aggregate demand. An important characteristic of these expenditures is that they represent consumer investments. Houses, vehicles, and household durables are purchases not just for current use, but to provide for future consumption. They are relatively infrequent transactions and often planned in advance. The asset value of these durables depends on the particular item, its condition, use and other economic factors, as well as on subjective considerations by the individual consumer. Replacement demand, for example, is more often to subjective valuation and preferences for upgrading, rather than whether the good remained in a usable condition. Since the average usable life of these physical assets is long, and the initial cost is high, such consumer purchases often involve the use of debt, which is to be repaid from expected future earnings.

CHART 1

TOTAL CAR, TOTAL TRUCK, AND TOTAL VEHICLE SALES

The third change in the macro economic environment involves the higher and more volatile rates of inflation experienced during the 1970s, and more recently, higher and more variable interest rates. From the mid 1950s to the mid 1960s, annual inflation rates were low and stable--averaging 1-2%. Since the mid 1960s, rates have been much higher and more volatile, ranging between 4% and 13%. Interest rates remained low and stable from the mid 1950s to the mid 1970s, with changes in the availability of credit the primary means to allocate credit. Since the mid 1970s, interest rates have increased substantially. For example, conventional mortgage interest rates remained at about 9% from 1974 through 1978, but rose in the following years to a peak of over 15% in 1982.

In the earlier time period, consumers reacted to inflation by decreasing spending and increasing saving, reflecting the short-term pessimism caused by higher inflation rates which were expected to fall in the future. Since the mid 1960s, and especially during the late 1970s, consumers reacted to rising inflation by increasing spending and decreasing saving. The buy-in-advance psychology which flourished in the late 1970s was overwhelmed in the early 1980s by reactions to lower inflation rates and rising interest rates. These changes in the aggregate economic environment increase the need for consumer decisions to take account of expected future developments over the near as well as longer terms in prices and interest rates.

CHART 2

NET ACQUISITION OF FINANCIAL ASSETS & LIABILITIES AS A PERCENT OF INCOME - 4 OUARTER MOVING AVERAGE

Spending and Saving Behavior

For the analysis of trends in aggregate demand, consumer outlays can be roughly categorized as either necessary or discretionary expenditures. The aggregate level. of expenditures for necessary budget items is more stable over time, and shows only small cyclical variation. Consumer decisions regarding necessary expenditures are frequently governed by need and guided by the force of habit; and in some cases, by contractual requirements--as for example, expenditures on shelter or loan repayments. Discretionary expenditures, in contrast, more frequently involve active decision making on the part of the consumer, and timing of the purchase is subject to greater variation. As a result, it is only for the analysis of cyclical variations in discretionary spending and saving behavior that consumer attitudes and expectations play an important role.

A major use of consumer attitudinal data is for the understanding and prediction of aggregate trends in consumer expenditures for housing, vehicles, and large household durables, as well as the incurrence of debt and the acquisition of financial assets. Katona postulated that economic optimism leads to a willingness to make large expenditures and debt commitments and that economic pessimism leads to a desire to curtail expenditures and rebuild financial reserves. When many people, at the same time, change from an optimistic to a pessimistic view of economic prospects, a subsequent and widespread shift towards postponement of discretionary expenditures follows.

TABLE 1

REGRESSION ANALYSIS OF TOTAL VEHICLE SALES IN THE U.S.  (1959:2 TO 1983:2, QUARTERLY DATA, SAAR)

This paper investigates two types of models in order to trace the relative impact of various measures of consumer sentiment: one focuses on trends in vehicle purchases. the other on the incurrence of debt by consumers. Chart 1 displays the time-series variations in total recall sales of all vehicles, including domestic and imported, as well as passenger cars and light-duty vans and trucks. Vehicle sales have shown greater cyclical variation during the decade of the 1970s compared with the 1950s and 1960s when secular increases dominated cyclical trends. The substantial impact on the aggregate economy from changes in vehicles sales is indicated not only by the rapid decline from the peak rate of 15.7 million units sold in the second quarter of 1978 to 9.9 million units sold just two years later, in the second quarter of 1980, but also from the persistence of low sales levels--two years after the initial decline, vehicle sales remained at 9.9 million units.

Chart 2 displays the trends in debt incurrence as well as acquisitions of financial assets by American households. All types of financial liabilities are included (mortgages, installment loans, credit cards, etc.) as well as all types of financial assets held by private households (checking and savings accounts, money market funds, bonds, stocks, etc.) The figures plotted are the proportion of current personal disposable income which is represented by net debt incurrence and net financial asset acquisition. Of course, the difference between these two percentages is the net household saving race. These data indicate that cyclical trends in the net saving rate are dominated by trends in debt incurrence, not asset acquisition. Consumer decisions about the use of debt are consequently a major determinant of the aggregate saving rate, and hence prospects for investment and economic (growth.

TABLE 2

REGRESSION ANALYSIS OF TOTAL CONSUMER DEBT INCURRENCE  (1959:2 TO 1983:2, QUARTERLY DATA, SAAR)

Expenditure Model

The basic model selected for the investigation of vehicle sales is widely used, and can be motivated, alternatively, by an underlying stock adjustment theory, by the permanent income hypothesis, or by psychological consideration. The basic equation, after suitable manipulation depending on the initial hypothesis, can be written as:

(1)    C_t = a + bC_t-1 + cY_t-1 + dS_t-1 + e_t

where:

C = total vehicle sales in units

Y = real personal disposable income per household

S = sentiment measures

Three separate equations were estimated, each including different measures of consumer sentiment. The measures of consumer sentiment selected for testing were:

ICS = Index of Consumer Sentiment

MKT = Evaluations of buying conditions for vehicles

TPR = Traditional price rationales given for market evaluations

ABR = Advance buying rationales given for market evaluations

The Index of Consumer Sentiment is a summary measure of trends in consumer attitudes and expectations derived from five regularly repeated questions, two on personal finances, two on business conditions, and one on buying conditions for large household durables. The second sentiment measure specifically asks consumers to rate buying conditions for vehicles. Since it's just as important to know why consumers feel as they do as it is to know how they feel, respondents are also asked to state their reasons for holding favorable or unfavorable views of market conditions for vehicles. The fixed-question free-response interviewing method provides detailed explanations that are recorded as completely as possible. The variable "traditional price rationales" reflects the proportion of consumers that cited the appeal of lower prices as the basis for their favorable attitudes, minus the proportion that cited high prices as the cause of their unfavorable attitudes. In addition to these traditional reactions to trends in vehicle prices, expected future trends in prices and interest rates have led consumers to buy in advance of the price increases, or to postpone purchases when interest rates are expected to decline. This variable is constructed by taking the proportion of consumers that cited buying-advance price rationales, minus the proportion that referred to high interest rates as a cause for purchase postponement. (For a detailed description of the sample, survey design, and question wording, see Curtin. 19825.

Because of the presence of a lagged dependent variable, estimates of the model's parameters by ordinary least squares would be inappropriate. In order to provide unbiased and consistent estimates, an instrumental variables approach was used, with an instrument for lagged vehicle sales entered in the second stave of estimation.

As shown in Table 1, the Index of Consumer Sentiment, as well as vehicle market evaluations, and the reasons underlying these evaluations proved to be significant predictors of vehicle sales. Comparing equations (1) and (2), the Index of Consumer Sentiment equation accounts for somewhat more variance in vehicle sales than attitudes toward market conditions. Equation (3), which includes both net price rationales, accounts for the largest amount of variance, outperforming both the more general index measure as well as the market evaluation question from which the price rationales were drawn. Both types of price rationales played an important role in shaping overall sales trends. The more traditional price rationales reflect consumer resistance to price increases, as well as the widespread appeal of rebates and price discounts. Advance buying rationales had a greater impact on vehicle sales-for each percentage point change in references to traditional price rationales, vehicle sales increased by 30,000 units, while vehicle sales increased by 70,000 units for each percentage point increase in advance buying rationales. As respondents have explained it, the high rate of vehicle sales in 1978 was due to the widespread expectation of price increases coupled with the belief that it was better to buy in advance of those increases, while at the same time interest rates were viewed as low and credit conditions easy. In contrast, the low sales rate recorded in 1981 and 1982 resulted from widespread postponement because of high interest rates, coupled with the view that prices were currently high and might fall later.

Debt Model

The model used to estimate trends in the net acquisition of financial liabilities as a percent of personal disposable income is similar to the expenditure model, and can be written as:

(2)    D_t = a + bD_t-1 + cA_t-1 + dS_t-1 + e_t

where:

D = Net change in debt as a percent of personal income.

A = Net change in financial assets as a percent of income.

S = Sentiment measures.

For the analysis of the debt variable, instead of using personal disposable income as a predictor variable, the net change in financial assets as a percent of personal disposable income was used. The ICS remains unchanged from the earlier model, but instead of just focusing on vehicle market evaluations, evaluations of the vehicle, house, and household durables markets were combined, as were references to the underlying price rationales.

As shown in Table 2, the Index of Consumer Sentiment was a significant predictor of debt trends, as were evaluations of overall buying conditions. In contrast to the vehicle demand model, the equation which included evaluations of buying conditions accounted for a somewhat higher proportion of variance than the Index of Consumer Sentiment. Equation 3, which included the price rationales, again outperformed both the Index and the overall market evaluations. Interestingly, only advance buying rationales significantly affected debt incurrence, not traditional price rationales, yet the equation still accounted for more total variance than the overall market evaluations from which these reasons were drawn. These results suggest that although traditional price rationales do affect vehicle purchase decisions, consumer decisions about the use of debt have been primarily affected by advance buying rationales.

The heightened sensitivity to expected trends in prices and interest rates led consumers to increasingly favor spending as well as the incurrence of debt during the late 1970s, while in the early 1980s, lower expected rates of inflation coupled with higher expected interest rates led consumers to postpone spending and decrease their use of debt.

Discussion

In the two decades following World War II, rising mass affluence led to an era of consumer optimism and confidence. These years were only briefly interrupted by periods of increasing pessimism, caused by concerns with income and employment prospects. During the past decade, rather than prosperity, slow and uneven economic growth has led to a new economic era which has been dominated by consumer pessimism and uncertainty. Consumer responses to economic adversity were not simply the reverse of their prior responses to prosperity. The escalating rates of inflation in the 1970s caused trends in consumer attitudes and expectations to diverge. As the expected rate of inflation increased, consumer expectations about their own financial situation as well as that for the economy as a whole declined in response, while the same price expectations led consumers to favor buy-in-advance rationales with regard to market conditions. Consequently, rising inflation both caused a sharp decline in evaluation of expected economic conditions, as well as helped maintain favorable evaluations of current buying conditions. These results indicate that consumers reactions to changes in their economic environment have become more complex during the past several decades, more dependent on longer term expectations, and if slow economic growth persists, more important for the analysis of cyclical trends in consumer spending and saving behavior.

REFERENCES

Curtin, Richard T., "Indicators of Consumer behavior: The University of Michigan Surveys of Consumers," Public Opinion Quarterly, Vol. 46, 340-352, 1982.

Katona, George. Psychological Economics. New York: Elsevier Scientific Publishing Company, Inc., 1975.

----------------------------------------