Estimation of past prices

9
Jourm~I uf Economic Psychology 8 (1987) 181-189 North-Holland ESTIMATION OF PAST PRICES Simon KEMP * University of Canterbury, Christchum& New Zeaiand Received October 31,1985; accepted March 13.1986 A survey of 271 Christchurch residents revealed that costs of the previous year were underesti- mated while those of 15 years previously were overestimated. These effects held for both estimates of costs in general and estimates of the costs of specific items. The results suggest that past costs are not remembered but estimated in a general way. Expectations of the extent of future price rises were related to estimated pas& price rises. Introduction A recent survey undertaken by Kemp (1984) showed New Zea- landers to underestimate systematically the extent to which general costs increased over the period 1967-1983. The underestimation was seen to be consistent with previous experiiental work showing sys- tematic misperception of exponential growth (Wagenaar 1982; Wagenaar and Sagaria 1975). Two issues left unaddressed by Kemp (1984) provided reasons for conducting the present study. Firstly, the previous study did not attempt to assess memory for the pact costs of specific items. Questions asked were of the type ‘Suppose you wish to buy something that costs $1,000 today. How much do you think it would have cost . . . years ago?‘, and respondentlr were specifically requested not to try to remem- ber what they paid for specific things. Yet how well people remember * I gratefully acknowledge the work of Phoebe Gray and Mary Harte, who carried out the interviews. I am also grateful for the suggestions and helpful advice of Barrie Stacey and the useful suggestions of Prof. WImeryd and Prof. Van Raaij. The research was supported by the New Zealand Dept. of Labour under the Project Employment Programme. Author’s address: S. Kemp, Dept. of Psychology, University of Canterbury, Private Bag, Christchurch, New Zealand. 0167-4870/87/$3.50 0 1987, Elsevier Science Publishers B.V. (North-Holland)

Transcript of Estimation of past prices

Jourm~I uf Economic Psychology 8 (1987) 181-189 North-Holland

ESTIMATION OF PAST PRICES

Simon KEMP * University of Canterbury, Christchum& New Zeaiand

Received October 31,1985; accepted March 13.1986

A survey of 271 Christchurch residents revealed that costs of the previous year were underesti- mated while those of 15 years previously were overestimated. These effects held for both estimates of costs in general and estimates of the costs of specific items. The results suggest that past costs are not remembered but estimated in a general way. Expectations of the extent of future price rises were related to estimated pas& price rises.

Introduction

A recent survey undertaken by Kemp (1984) showed New Zea- landers to underestimate systematically the extent to which general costs increased over the period 1967-1983. The underestimation was seen to be consistent with previous experiiental work showing sys- tematic misperception of exponential growth (Wagenaar 1982; Wagenaar and Sagaria 1975).

Two issues left unaddressed by Kemp (1984) provided reasons for conducting the present study. Firstly, the previous study did not attempt to assess memory for the pact costs of specific items. Questions asked were of the type ‘Suppose you wish to buy something that costs $1,000 today. How much do you think it would have cost . . . years ago?‘, and respondentlr were specifically requested not to try to remem- ber what they paid for specific things. Yet how well people remember

* I gratefully acknowledge the work of Phoebe Gray and Mary Harte, who carried out the interviews. I am also grateful for the suggestions and helpful advice of Barrie Stacey and the useful suggestions of Prof. WImeryd and Prof. Van Raaij. The research was supported by the New Zealand Dept. of Labour under the Project Employment Programme.

Author’s address: S. Kemp, Dept. of Psychology, University of Canterbury, Private Bag, Christchurch, New Zealand.

0167-4870/87/$3.50 0 1987, Elsevier Science Publishers B.V. (North-Holland)

the past cost of specific items may be important for determining the process responsible for the underestimation of costs in general. Sec- ondly, the earlier work did not explore people’s expectations of future changes in the cost of living. It would, however, be interesting to discover if those people who underestimate past cost &o expect little change in the future.

271 respondents were interviewed between May and August of 1985. Half the respondents were tained by area sampling and interviewed in their homes; the other half were obtained from a selection of public places and interviewed at these places.

The persond information collected suggests the sample to be rea- sonably representative of the ad& Christchurch population. The full sample contained 106 men and 165 women, and ranged in age from 18 t3 90. Information on the respondents’ socioeconomic status and political party support showed no apprecipble biases.

All respondents were interviewed by one of two paid interviewers using an interview schedule. Respondents were not permitted to I&W the schedule while being interviewed.

l%e intemiew

All interviews followed a set format determined by the structure of the schedule. The majority of the questions covered either general costs or the cost of specific items. Half the respondents were interviewed from schedules in which the general cost questions preceded those relating to specific items, half from schedules in which the general cost questions followed those relating to specific items. Some requests for personal information concluded the schedule.

General cost questions were of the form:

‘Suppose you wish to buy something that costs ten dollars today. How much do you think it would have cost one year ago, in 19&i?’

S. Kemp / Estimation of pat prices 183

Two similar general questions asked the cost of something fifteen years ago (1970) and in five years’ time (1 . Responden& were

ins&ucted not to think of particular items while answering the general questions.

Specific item questions were of the form:

‘How much do you think the stamp for a standard surface mail letter to a N.Z. address costs today?’ ‘How much do you think 500 g of butter costs today? ‘How much do you think the telephone bill (excluding tolls) costs every two months?’ *

Three further questions about each of the stamp, butter, and tde- phone items asked the cost of the item one year ago, fifteen years ago, and in five years’ time. All questions relating to one item were corn- pleted before the next item was introduced. The choice of the items was determined by considerations of familiarity to the respondents and the availability of accurate records on their past prices. Respondents were also asked whether they did most, some, or none of the grocery shopping for their household, and whether they personally paid the telephone bill for the household.

Respondents were specifically requested to guess when they did not know or could not remember the prices, and were asked for ‘quick, rough estimates’.

Results

Table 1 shows the median estimates of the respondents in response to the genera’& stamp, butter, and telephone questions, and the corre- sponding actual prices. For general costs the actual prices were ob- tained by averaging the Consumer Price indices for the June and September quarters of the appropriate years (fiepartment of Statistics 1970-1985a). For the specific items the prices were obtained by averag- ing the surveyed prices in the months May to August (Department of

Statistics 1970-1985b). As table 1 shows, estimates of 1985 prices were quite accurate, 1984 prices underestimated, and 1970 prices overesti

1 In New Zealand telephone accounts are invoiced every two months. Local calls are not charged separately, so the bill covers all but long-distance or toi;. cal!s.

184 S. Kemp / Estimatkm of past prikes

Table 1 The upper y._ttion of the table shows the median estimates (with the lower and upper quartile estimates in parentheses of the cost of a general $10 item, a stamp, 500 g of butter, and the telephone bill at the time of the survey (1985), one year ago (1984), fifteen years ago (1970), or in five years* time (1990). The lower portion shcws the actual costs dcrkd from official figures (Department of Statistics 1970-198% b).

Year General ($0 Stamp (h Butter (cc> Telephone ($)

Estimated costs

1985 - 24 (24,25) 139 (125,157) 33.60 (33,38) 1984 8.00 (7,8.50) 24 (20,24) 100 (95,119) 32.00 (27,33.20) 1970 4.00 (2.50,5) 6 (4910) 50 (30,60) 16.08 (10,22) 1990 16.08 (1520) 35 (30,50) 200 (190,250) 45.00 (40,60)

Actunl costs

1985 - 24 139 34.01 1984 8.59 24 115 33.20 1970 1.73 3 30 7.33

mated. The overestimation of 1970 prices is particularly striking in table 1 and clearly common to both prices in general and prices of the specific items.

Table 2 shows the correlation between estimated increases in general costs and estimated increases in the costs of the specific items. Esti- mated increases in costs of specific items were obtained bj, dividing the respondent’s estimated past or future cost by the respondent’s estimate of the present cost of the item. The positive correlations shown in table 2 indicate that people who estimated a relatively large price

Table 2 C&relation coefficients (r) between respondents’ estimates of general cost increases and estimates of price increases of specific items (stamp, butter, and telephone) since 1984 and 1970 and (predicted) for 1990. The number of respondents on which each r is based is shown in parentheses.

Y&U

1984

1970

1990

ilswsm-,sm~

Stamp

0.32 (269) 0.22 (265) 0.56 (268)

Butter Telephone

0.32 0.11

(262) (258) 0.20 0.38 (260) (249) 0.39 0.57 (259) (263)

S Kemp / Estimation of past prices 185

increase on one item also estimated a relatively large increase in genead

costs.

An obvious interpretation of the results shown in tables 1 and 2 is that respondents remembered or misremembered the prices of specific items during the interview and then ‘averaged’ these estimates to obtain an estimate of c2neraI costs. This interpretation, however, must be rejected as no significant difference in any of the three general cost estimates was found between those respondents who were questioned on general costs before being questioned on specific items and those who were questioned on general costs afterwards (1984 estimates, F(l, 268) = 0.14, n.s.; 1970 estimates, F(l, 267) = 0.08, n.s.; 1990 estimates, F(l, 267) = 2.26, n.s.).

Table 3 shows the correlations between the estimates of past in- creases and the predicted price increases over the next 5 years. The moderate correlations displayed between the past and future price estimates indicate that people who perceived lower price increases over the previous l- or U-year period also expected a lower rate of price increase in the future. Thus individual expectations of future price changes were related to individuaI misperceptions of changes in past prices.

Analyses of variance revealed no significant ( p < 0.01) effects of political party supported, respondent’s socioeconomic status, marital status, or extent of personal impact of inflation on any cost estimate. The age and sex of the respondent were found to be significantly related to estimates of stamp (age, F(1, 263) = 14.42, p < 0.01; sex, F( 1, 263) = 7.90, p < 0.01) and telephone (age, F(l, 247) = 10.80, p < 0.01; sex, F(l, 247) = 7.80, p < 0.01) costs in 1970. In both cases

Table 3 Correlation between pre.di%d future price increases and estimated past price increases. Correla- tion coefficients (r) between respondents’ estimates of cost increases over the next 5 years and their estimates of cost increases over the past 1 and 15 years are shown for each specific item as well as general costs. The number of respondents on which each .* is based is shown in parentheses.

General stamp Butter Telephone

Past 1 year 0.19 0.14 0.08

(269) (260) (255) ;z, Past 15 years 0.29 0.16 0.19 0.26

(268) (260) (255) (241)

186 S. Ken@ / Estimation of past prim

men estimated lower (and more accurate) prices than women, and respo&nts aged under 30 estimated lower (and more accurate) prices than those over 30. There were no further effects of age and sex on

pia e&mates of specific items but there were significant effects of age and sex on perceived general costs in 1984 (age, F(1, 268) = 19.56, p < 0.01; sex, F(1,268) = 31.72, p < 0.01). Men gave higher (and more accurate) general cost estimates than women, while respondents aged under 30 gave hi&~ (and more accurate) estimates than those over 30.2 There were no significant effects on the 1970 or 1990 general cost estimates. Overall, particularly when it is considered that none of the predictor variables studied related to any future price estimates, it appears that perception of price rises is comparatively unaffected by attitudinal and demographic variables.

Respondents’ estimates of future and present butter prices and their estimates of the 1984 price of butter were unrelated to whether they reported doing moat, some, or none of the household grocery shopping. On the other hand, the proportion of shopping done bore a significant relationship to estimates of the 1970 butter price (F(2, 259) = 6.55, p < 0.05). Avemge but&r prices estimated were 56$ by those who reported doing most of the grocery shopping, 46# by those doing some, and 4l# by those doing none. Thus the most accurate 1970 estimates were given by those who did not buy the butter. SimGrly, no signifi- cant differences in telephone bill estimates for one year i go, five years’ time, or the present were found between respondents who reported paying and not paying the bill, while non-payers gave a lower and more accurate average estimate ($14.91) than payers ($17.40) for the bill of fifteen years ago (F(1, 246) = 5.51, p < 0.05).

Probably as a consequence of the instructions given to respondents, there were few misskg data (‘don’t know’) responses for any question, the most being 22 for the 1970 telephone question. Examination of. the missing data revealed no clear patterns: for example, younger respon- dents did not produce more missing data in response to 1970 questions than older respondents, although clearly they must have estimated rather than remembered their responses.

One further result deserves mention. Typically in New Zealand, the price of a standard stamp has remairzd unchanged over fairly long

* Remember that 1984 prices were underestimated while 1970 prices were overestimated (see table 1).

S. Kemp / Estimation of part prices 181

periods, and then risen sharply overnight (e.g., from 4# to 84 in 1976). In fact 105 of the 265 respondents estimated a 1970 stamp price (e.g., 5d) that never existed at any time.

The results of the present study replicate the finding of Kemp (1984) that past prices tend to be overestimated and extend the finding to specific items. Table 1 indicates that the prices of even common, everyday items are erroneously recalled or misperceived in just the same way as the trend of prices in general. This misperception occurs despite the fact that present prices of items were generally accurately estimated, implying that, for example, prices of items 15 years ago were also known accurately at the time.

The general pattern of results outlined in the previous section is clearly complex. Many of the fkdings, however, can be summarised by three principles.

Firstly, in general, people do not really ‘remember’ the past prices of specific items at all, particularly when they are asked for 15-year-old prices. This is not a surprising result: research into very long-term memory has shown that people of all ages are frequently unable to recall details of well-known events a few years after the events occurred (Squire and Slater 1975; Warrington and Sanders 1971; Warrington and Silberstein 1970). Jenkins et al. (1979) recently reported that subjects who were asked to record important life change events over a six-month period forgot nearly half of them when asked to recall them six months later. Against this background it is unsurprising that subjects in the present study could not recall, for example, the price they had paid for stamps 15 years previously.

Secondly, respondents in this survey appeared to use a general model of how costs change over time which they applied both to specific items and costs in general. Evidence for this model is the positive corre1ation.s shown in tables 2 and 3, indicating that individuals perceiving, say, a high rate of price increase did so generally and not just for a particular item or period. Systematic misperception of past prices seems to be a consequence of using the general model: respondents consistedy overestimated 1970 prices and underestimated 198 suggested that inaccurate estimation of past prices resulted from &en-

188 S. Kemp / Estimation of past prices

eral underestimation of the rate of exponential growth. Such a hypothe- sis predicts the 1970 price overestimation in the present study but does not, however, predict the 1984 price underestimation. Another possi- baty is that misperception of past prices in this study results from systematic distortion of perceived duration. Ferguson and Martin (1983), for exampaw, te found the age of recent (9-12 months) news items was overestimated while that of older news items was underestimated. Indeed, there appears to be a general tendency for stimuli to be remembered as more similar than they in fact are: short remembered distances are overestimated while longer ones are underestimated (Wiest and Bell 1985); similar results have been found for visual area (Kerst and Howard 1978).

Thirdly, familiarity with the prices of a specific item generally seems to have resulted in less accurate estimation of its past price. This result was clearly seen in the effects of paying the telephone bill and doing the grocery shopping. It is also consistent with the age and sex effects

- reported above. At first sight this general result is surprising but it is consistent with the principle of interference (see e.g., Woodworth and Schlosberg 1955: 733-778). According to this principle, recent prices will interfere with (and tend to replace) memory for more distant prices for the same item. Clearly this effect should be s:rongest for respon- dents who have most experience with the recent prices, i.e., those who do the gocery shopping or pay the telephone bill. An additional factor is that they were perhaps more likely to try to remember than to estimate past prices; similarly, people under 30 were perhaps less likely to believe they could remember I970 telephone bills they had no experience of.

One final result of this study me& further discussion. Previous research (e.g., Blomyvist 1983; Ghan-Lee 1980; Jonung 1981) has considered the relationship between inflationary expectations and. ac- tual or estimated present inflation rates. The present study show? a moderate correlation between expected price changes and mispercep- tions of past price changes. This correlation holds not only for prices in general, where we might expect indiviouai difkences to reflect differ- ent purchases of goods and servicep, but also for the prices of specific items which are generally not different for different individuals. Thus it appears that inflationary expectations might depend on perceptual and cognitive factors as well as on real changes in prices.

S. Kemp / Estimation of part prices 189

References

Blomqvist, H.C., 1983. On the formation of inflationary expectations: some empirical evidence from Finland. Journal of Economic Psychology 4,319-341.

Ghan-Lee, J.H., 1980. A review of recent work in the area of inflationary expectations. Weit- wirtschaftliches Arcbiv 116,45-86.

Department of Statistics, 1970-1985a. New Zealand monthly abstract of statistics. Wellington: Government Printing Office.

Department of Statistics, 1970-1985b. New Zealand statistics: prices, wages and labour. Welling- ton: Government Printing Office.

Ferguson, RP. and P. Martin, 1983. Long-term temporal estimation in humans. Perception and Psychophysics 33,585-592.

Jenkins, C.D., M.W. Hurst and RM. Rose, 1979. Life changes: do people really remember? Archives of General Psychiatry 36,379-384.

Jonung, L., 1981. Perceived and expected rates of inflation in Sweden. The American Economic Review 71,%1-968.

Kemp, S., 1984. Perception of changes in the cost of living. Journal of Economic Psychology 5, 313-323.

Kerst, S.M. and J.H. Howard, 1978. Memory psychophysics for visual area and length. Memory and Cog&ion 6,327-335.

Squire, L.R and P.C. Slater, 1975. Forgetting in very long-term memory as assessed by an improved questionnaire technique. Journal of Experimental Psychology: Human Learning and Memory 1,50-54.

Wagenaar, W-A., 1982. ‘Misperception of exponential growth and the psychological magnitude of numbers’. In : B. Wegener (ed.), Social attitudes and psycho-physical measurement. Hillsdale, NJ: Erlbaum. pp. 283-301.

Wagenaar, W.A. and S. Sagaria, 1975. Misperception of exponential gtowth. Perception and Psychophysics 18,416-422.

Warrington, E.K. and HI. Sanders, 1971. The fate of old memories. Quarterly Journal of Experimental Psychology 23,432-442.

Warrington, E.K. and M. Silberstein, 1970. A questionnaire technique for investigating very long-term memory. Quarterly Journal of Experimental ?sychology 22,508-512.

Wiest, W.M. and B. Bell, 1985. Stevens’s exponent for psychophysical scaling of perceived, remembered, and inferred distance. Psychological Bulletin 98,457-470.

Woodworth, R.S. and H. Schlosberg, 1955. Experimental psychology. London: Methueu.