david leiser* corresponding author: david leiser, dept of
TRANSCRIPT
Lay understanding of macro-economics. 1
Lay understanding of macroeconomic causation
the good-begets-good heuristic
David LEISER*
Ronen AROCH
Ben Gurion University
Corresponding Author:
David Leiser, Dept of Behavioral Sciences, Ben-Gurion University, P.O.Box 653, Beer Sheva
84105, Israel
Phone: + 972 544979225
Fax: + 97225618841
Email: [email protected]
Submitted for publication 2007
Lay understanding of macro-economics. 2
Abstract
The functioning of the economic system is extremely complex and technical. And yet the
public is constantly presented with information on economic causality. It is important for its
members to assimilate it, be it to further their personal goals or in order to participate
advisedly in the democratic process.
We presented naïve and economically educated participants with questions of the form:
If VARIABLE A increases, how will this affect VARIABLE B ? for all possible combinations of 19
key economic indicators. Naïve participants were willing to commit themselves on most
questions, despite their medium to low self-report of understanding the concept involved.
Analysis of the pattern of responses reveals the use of a simple heuristic: good-begets-good
that yields a sense of competence without requiring understanding of the causality involved.
Lay understanding of macro-economics. 3
Authors’ note
*Address all correspondence to David Leiser, [email protected]. We thank Diana Rubin for
her pioneer work and Channa-Muriel Setbon and Ofer Azar for their professional help.
Lay understanding of macro-economics. 4
In democracies, policies are heavily influenced by the views and beliefs of the public. This
raises a problem when the beliefs of the public are at variance with those of specialists. How
can a democracy function, it may be asked, when the public do not understand the issues at
stake? Two domains where these questions have been raised are politics and economics
(Carpini & Keeter, 1997; Gilens, 2005; van Bavel & Gaskell, 2004). In particular, the
importance of economics for elections has been noted (Lewis-Beck, 1988; Lewis-Beck &
Rice, 1992). Individual attitudes, preferences and behaviors are moderated by levels of
factual information about relevant stimulus objects (Hausman, 1992).
In the domain of economics that will concern us here, differences between the public
and professional economists have been noted. (Blendon et al., 1997) report the results of two
surveys, one of economists, one of the public and finds that the public has a bleaker picture of
what has happened economically to the average family, is more pessimistic than most
economists about the intermediate future. The public also cites different reasons than
economists do for why the economy is not doing better. Economic beliefs of economists and
the public differ systematically (Caplan, 2001; Walstad & Rebeck, 2002), with factors such
as education, income changes or job security affecting the likeness of their outlook. That fit
is not the result of a self-serving bias, since materially advantaged noneconomists think much
like the rest of the public, not like economists (Caplan, 2002). Economic training, as may be
expected, is a significant factor. The differences examined in these studies are about the
relative importance ascribed by the respondents to economic factors. Economists, for
instance, consider foreign trade and downsizing as helpful, accept supply and demand
explanations rather than monopolistic explanations of price changes.
A different approach to the study of lay understanding of economics relies on the
social representations approach to discover relations between concepts. Social representations
are defined as socially shared ideas, opinions, attitudes, and theories (Moscovici, 1981,
Lay understanding of macro-economics. 5
1984). Social representations are influenced by public discourse (mass media, newspapers,
school, the discourse of politicians and so forth), yet also have their own coherence. They
affect the diffusion of normative theories used by professionals, assimilating but also
distorting and modulating them. In practice, a common method consists in presenting to the
participants a term and to see what others are evoked. Based on the pattern of concepts
evoked (order, frequencies) it is possible to deduce the underlying social representations
(Leiser & Drori, 2005; Verges, 1989).
The present study will take a different tack, and examine the causal network formed
by economic concepts, according to laypeople, and compare it with that of economist. Thus
we will study the existence and direction of causal relations between pairs of variables, not
their relative importance.1.
In a preliminary study, we presented students with about 20 economics variables, such
as the economic growth, inflation rate, savings rate, and asked them to form as many pairs as
they wanted, such that one of the variables causes an increase or a decrease in the other. For
example, a subject might judge that an increase in income would provoke an increase in
consumption, and in savings. Such directed relations can then be combined in a causal map
(Axelrod, 1976; Eden, Ackermann, & Cropper, 1992; Laukkanen, 1994, 1996a, 1996b). The
resulting maps are complex. Figure 1 shows the network of relations expressed by one of the
participants in that study.
Insert Figure 1
1 Causal relations, especially in economics, do not form a network of simultaneous relations. The causal system
plays out over time. However, we will ignore the temporal dimension in this study..
Lay understanding of macro-economics. 6
The psychological structuring of causality in macro- economics
Macroeconomics studies the economic functioning of a country as whole, what is
often called the behavior of the aggregate economy. The markets for goods and for services,
the work, financial and currency markets) are all interrelated. Macroeconomics examines
economy-wide phenomena such as changes in unemployment, inflation and price levels. To
evaluate the global performance of an economy, aggregate indexes are used: the GDP and its
growth rate, the inflation rate, the unemployment level, and so on. Economic science
develops models of the interactions between theses indexes. Those models are complex, and
often rely on the concept of equilibrium. Consider a short description of the supply-side
model.
Supply-side economists hold that if people had more cash in hand, they would spend
more on goods and services, thereby increasing the aggregate demand for those goods
and services and stimulating economic growth. Since there are natural limits to the
amount of goods and services people need, they would invest their surplus assets in
interest- or dividend-bearing securities, thus making more capital available for
investment and further driving down interest rates. Lower interest rates, coupled with
higher aggregate demand, would prompt businesses to borrow to fund expansion, a
development that also quickens economic growth (www.answers.com).
While such arguments are technical, the discourse on economics is not restricted to specialists
only, as is that of engineers, for instance. They are often explained in pieces intended for the
public. One can read them in mass-circulation newspapers, where the evolution of the stock
market or the rationale for the decisions of the central banks are discussed. The public can
also read popular columns with titles such as “how to invest your money?” which typically
contain a mix of analysis and advice.
What do laypeople make out of such discourse? In view of the complexity of the
causal relations in economics, it seems doubtful that they really understand it. As Arthur
(2000) stresses: economics is inherently difficult. If they try to make sense of it nonetheless,
Lay understanding of macro-economics. 7
they must impose some simpler structure or rely on heuristics. This is the claim we will try to
support.
Method
Participants
The main experiment included two groups. The first consisted of economically naïve
participants, who had no previous formal exposure to economic theory. Forty-two first year
students in psychology engaged in the experiment in the first group, as part of their course
duties. The second group consisted of economically educated participants and was composed
of eighteen students in their last year of bachelors’ degree in economics.
Procedure
The participants responded to a self-paced computer administered questionnaire in
individual booths. The main part of the questionnaire consisted in judgments on the causal
relationships between pairs of economic values. We selected 19 central economic variables,
based on a pilot study and the advice of an economist (see Table 1).
A fixed format was used for all the questions.
If VARIABLE A increases, how will this affect VARIABLE B ?
For example: If the unemployment rate increases how will this affect the inflation rate?
The participant had to select one answer out of four: (1) B will increase; (2) B will decrease;
(3) B will not be affected; and (4) “don’t know”. We avoided a forced-choice paradigm since
encouraging guessing amongst the truly ignorant biases estimates of knowledge for the
sample as a whole and reduce the reliability of summated scales., while Sturgis, Allum,
Smith, & Woods (2005) showed that responses given to requests for a ‘best guess’ after an
initial ‘don’t know’ fare no better than the expectation under a ‘blind-guessing’ strategy.
This format was used for all the possible combinations of variables in both directions, 342
questions in all. The order of the questions was determined at random for each subject.
Lay understanding of macro-economics. 8
Following this part, we asked 19 questions on their own understanding of the
variables involved. These questions formed part of the computer-administered session. The
phrasing was: To what extent to you understand the meaning of the following concepts.
Participants had to type a number (1-5) with the Likert scale materialized as a line with five
marked segments. The ends were labeled as: “1 – I do not understand it at all” and “5 – I
understand it very well.” Table 1 gives the mean values, and is ranked by decreasing
understanding of the naives.
Good/bad evaluation.
In the course of the analysis, it seemed instructive to obtain evaluations about what
changes in economic variables were considered good or bad. We recruited participants by
means of the internet, and selected those respondents who reported either having had no
formal economic training even at high school (N=85) or who had had economic classes in
college (N=47). Following the demographic data collection, these participants were asked:
Please mark for each of the following quantities whether an increase is a positive or a
negative development. Their answers were recorded on a 5 point Likert scale, with the
endpoints labeled Very good/Very bad, and the midpoint labeled Neutral.
Lay understanding of macro-economics. 9
Results
Table 1 - understanding of concepts - self-report
Variable Naives Economists
Rate of unemployment 4.67 4.39
Average net salary 4.33 4.22
Preference for local products 4.19 4.00
State welfare expenses 4.14 4.17
Profits by business sector 4.14 4.06
Loans taken by the public 3.83 4.11
Economy growth rate 3.79 4.28
Competitiveness in the market 3.71 4.06
Investment by the public in stock market 3.57 3.72
Income tax rate 3.52 3.78
Consumption rate 3.48 4.39
State Expenditures 3.43 4.50
Depth of recession 3.40 4.06
Quantity of money in the economy 3.31 3.89
Gross National Product 3.26 4.39
Inflation rate 3.12 4.22
Interest rate on loans 2.48 4.28
Saving rate 2.40 4.17
National credit rating 2.33 3.61
As expected, the mean degree of self-reported understanding is lower for the
economically naïve: (mean Naïve: 3.53; mean Economists: 4.11; t(18)=- 4.01; p=0.0008).
However, since the question was asked at the end of the questionnaire, there was reason to
fear that participants would be embarrassed to admit they knew little about the concepts they
had been answering questions about. We took therefore a new and comparable group of
subjects (N=36) from third year students in psychology and asked them merely to report, on a
Lay understanding of macro-economics. 10
pen and pencil sheet, how well they understood the concepts involved. The mean reported
degree of understanding dropped further, from 3.53 to 2.69 (t(18)= -3.83; p=0.001).
Table 2- Answering directional patterns
Answer type on couple proportion
Both positive 0.15
One positive one negative 0.09
Both negative 0.12
One direction 0.34
Neither 0.29
The “don’t know” rate is, on average, of 27%. In other terms, on 83% of the cases
they feel confident enough to answer. The distribution of their answers is interesting. For any
pairs of variables, A and B, the link between them can be both negative, both positive, one
positive and the other negative, one directional, or the participant may consider that the two
concepts do not affect one another. These various possibilities are presented in Table 2. As
may be seen, for any two variables, more than two third are considered to be related. When
both directions (A affects B and B affects A) are affirmed, they take the same direction (both
positive or negative) three times more often than opposite directions.
To obtain a synoptic view of the way variables are causally related according to the
participants, we used the number and direction of causal links between two variables as a
measure of their similarity. Whenever a participant affirmed that A increases B or that B
increases A, we added 1 unit to the relation between A and B. If the participant affirmed both,
we added 2 units. Similarly, whenever a participant affirmed that an increase in A decreases
B or conversely, we subtracted 1 unit. We then subjected the resulting triangular matrix to a
Lay understanding of macro-economics. 11
two-dimensional scaling procedure, using the MDS module in Statistica, from StatSoft
(Purkhardt & Stockdale, 1993). Figure 2 displays the resulting map.
Insert Figure 2 about here
The concepts are clearly clustered in two groups, separated by large divide. K-means
analysis yielded the clusters listed in Table 3.
Table 3 - Members of the two clusters
Positive pole Negative pole
GNP State Expenditures
National credit rating Inflation rate
Profits by business sector Income tax rate
Investment by the public in stock market . Interest rate on loans
Average net salary Rate of unemployment
Consumption rate Loans taken by the public
Quantity of money in the economy Depth of recession
Economy growth rate State welfare expenses
Competitiveness in the market
Preference for local products
Saving rate
A glance at the two clusters suggests an obvious interpretation: participants organize
the variables in two poles, one negative and the other positive. To bolster this interpretation,
we obtained goodness/badness judgments on changes in the values of variables from an
additional group of eighty-five participants (see the methods section above). The correlation
between the one-dimensional MDS and the good-bad dimension is r= .9346 (p<.0001).
In practice, when asked: does A influence B (and in what direction)? participants can
answer readily: if A and B belong to the same pole, a increase in one will raise the other; if
they belong to opposite poles, a raise in one causes the other to drop. This interpretation
explains the otherwise surprising willingness of participants to commit themselves on issues
Lay understanding of macro-economics. 12
that, by their own account, they do not understand very well. It is reinforced by the
correlation between how markedly positive or negative the concept is felt to be (as measured
by a one dimensional MDS) and the willingness to commit to an answer (proportion of
answers 1-3). That correlation is r=.44 (p=.057 two-tailed).
Do naives resemble economists more when the concepts involved are clearly good or
bad? First, we devised a measure of fit. Participants had four answering options for every
question in our questionnaire: If variable A increases, then variable B will ___: increase,
decrease, no effect, and DK. Leaving aside the DK answer to which we return below, we
tabulated the distribution among the three substantive answers on every question for each
group (naïve and economists) separately. We then tested for each question whether the
distribution differed between the groups at the α=.05 level, by running a chi-square test.2
Thus, we could determine for every question whether the two groups gave the same pattern of
answers.
We then categorized the questions according to whether the good/bad heuristic could
be used. Since answer on the good/bad dimension were given on a bi-polar scale (Bad –
Neutral – Good) we classified each variable according to the nearest of these three points.
That is, if the mean answer of the naïve subjects, for a given variable was closer to Good than
to Neutral, it was counted as Good, as Neutral otherwise; the equivalent was done at the
negative end of the scale. This divided all concepts in 3 groups, Good, Neutral, Bad, which in
turn yielded 9 combinations for the pairs of variables Pairs with at least one neutral concept
do not allow the good/bad heuristic, all the others combinations do. When the heuristic is
2 Overall, we found a 62% fit, that is, for 62% of the questions, no significant difference was
found between the groups.
Lay understanding of macro-economics. 13
relevant, there was a .71 proportion of fit between naives and economists. When it wasn’t, it
dropped to .31.
This result suggests that economists reflect the good/bad dimension is their thinking.
We generated a MDS plot (see Figure 3) by the same procedure as used for the naives. As
may be seen, their maps are indeed quite similar.
How accurate is the Feeling of Understanding of the naive participants?
First, we tested whether the “don’t know” (DK) response is resorted to more often
when one of the concepts involved is less familiar. As expected, participants are more to take
a stand on the relation of a concepts to others when they understand it better (correlation
between self-reported understanding and us of DK: r=-.72; p<.0001). The correlation for
economists goes in the same direction, but there is a ceiling effect and the effect is not
significant (r=-.34, p=.154). This merely shows that the participants are self-consistent. A
more important question is: is their self-estimate of understanding valid? When they feel
they understand the concepts involved, do their judgments converge with those of
economists?
We described above how we determined for every question whether the two groups,
naive and economists gave a significantly different pattern of answers or not. When the
difference was not significant, we considered the two groups were matched in their answers.
We now ran a probit analysis, with the self-reported extent of understanding of the antecedent
(A) and the consequent (B) variables involved as predictors, and whether there was a match
between the groups as the predicted variable. Both predictors proved highly significant, with
the effect of the antecedent variable much larger than that of the consequence (Wald
coefficient antecedent: 34.36, p<.0000001; Wald coefficient consequence: 14.57, p<.0001)
(an asymmetry that may be related to Sevón’s (1984) observation of an easier flow from
cause to effect than conversely). We also tested whether the good/bad dimension improved
Lay understanding of macro-economics. 14
the prediction. Since the good/bad dimension GB was measured on a 5 point scale, we
introduced as predictor variable abs(GB-3) besides the previous two factors. This new factor
did not approach significance, and did not contribute at all to a prediction of the match
between naïve and economists beyond self-reported understanding.
Conclusions
We set out to understand how economically naive people handle economic causal
discourse to which they are constantly exposed. Their answers are at first sight paradoxical:
on the one hand, they declare on average not to understand the concepts very well. On the
other, they are willing to judge how changes in one economic variable would affect another.
Our interpretation is that what enables the naives to answer is their superficial approach to the
issues. The domain of economic variables is for them bipolar, a tendency identified by Brown
(1991) as a universal tendency of human nature. Economic events are classified as good or
bad, not as neutral components in a causal system: "Social actors do not try to isolate an
economic object from social reality; on the contrary, they relate social and economic
elements in their representation (Vergès,1989)."
We submit that naïve participants rely on a simple but powerful heuristic: the
economic world functions in either a virtuous or a vicious circle. An increase in one good
variable will increase the values of other good variables, and decrease those of bad variables.
This good-begets-good heuristic settles in most cases how to answer. It is not unrelated to
the way economic events are commonly described in popular economic discourse, with
strong valuation of every change as either positive or negative. Here is a sample at random:
The day's news is mostly positive. The better than expected housing starts numbers
come a day after the National Association of Homebuilders Sentiment Index [HMI]
rose more than expected. It seems now that the worries about a housing debacle in
2007 might have been exaggerated. As a result, stocks are likely to respond favorably
Lay understanding of macro-economics. 15
to today's numbers because a period of improving economic growth and low inflation
is ideal for corporate earnings3.
Professional economists are certainly not foreign to looking at the economy as being
in a good or bad state. Indeed, they have long devised economic indices to indicate an overall
evaluation of the state of the economy. The first “misery index”, introduced by Arthur Okun
in the Lyndon Johnson years, is simply the unemployment rate added to the inflation rate.
Later, Robert J. Barro (1999) refined the misery index by adding gross domestic product and
interest rates to inflation and unemployment. Recently, Merrill Lynch's economists devised a
yet broader index (see The Economist, Jan 12th 2006): it adds unemployment and inflation
rates, interest rates and the budget and current-account balances, but then subtracts GDP
growth. Its rationale is that high unemployment, inflation, and interest rates are bad, whereas
positive budget and current account balances and a high GDP growth rate are good.
Here the naïve participants show commendable restraint. They do not understand all
concepts equally well, avoid committing themselves when the meeting of the quantities
become more obscure to them, and are mostly right in their self-evaluation: the better they
feel they understand a concept, the more often their judgment will match those of the
economists.
The good-begets-good heuristic does not imply the ability to explain causal links. The
depth of explanation of economic concepts is low in all segments of the population (Leiser &
Drori, 2005), and indeed, the lack of explanatory depth seems to be a general feature of
human understanding (Keil, 2003; Keil, 2006; Leiser, 2001; Rozenblit & Keil, 2002).
3 http://www.insidefutures.com/article/3641/MORNING%20WATCH,%20Jan.%2018.html
Lay understanding of macro-economics. 16
We set out to describe how the public handles economic theory thrust in its face,
when it is patently far more complex than it can handle. The answer is: the combination of a
useful heuristic, reasonable self-evaluation and shallow understanding.
Lay understanding of macro-economics. 17
References
Arthur, W. B. (2000). Cognition: The Black Box of Economics. In D. Colander (Ed.), The
Complexity Vision and the Teaching of Economics, . Northampton, Mass: Edward
Elgar.
Axelrod, R. (1976). The analysis of cognitive maps. In R. Axelrod (Ed.), Structure of
Decision: The Cognitive Maps of Political Elites (pp. 55-72). Princeton, NJ: Princeton
University Press.
Barro, R. J. (1999). Reagan vs. Clinton: who's the economic champ? Business Week, 22.
Blendon, R. J., Benson, J. M., Brodie, M., Morin, R., Altman, D. E., Gitterman, D., et al.
(1997). Bridging the Gap Between the Public's and Economists' Views of the
Economy. The Journal of Economic Perspectives, 11(3), 105-118.
Brown, D. E. (1991). Human Universals. New York: McGraw-Hill.
Caplan, B. (2001). What Makes People Think Like Economists? Evidence on Economic
Cognition from the“Survey of Americans and Economists on the Economy”. The
Journal of Law and Economics, 44(2), 395-426.
Caplan, B. (2002). Systematically biased beliefs about economics: Robust evidence of
judgemental anomalies from the survey of americans and economists on the economy.
The Economic Journal, 112(479), 433-458.
Carpini, M. X. D., & Keeter, S. (1997). What Americans Know About Politics and Why It
Matters: Yale University Press.
Eden, C., Ackermann, F., & Cropper, S. (1992). The Analysis of Cause Maps. Journal of
Management Studies, 29, 309-324.
Gilens, M. (2005). Political Ignorance and Collective Policy Preferences. American Political
Science Review, 95(02), 379-396.
Lay understanding of macro-economics. 18
Hausman, D. M. (1992). The Inexact and Separate Science of Economics: Cambridge
University Press.
Keil, F. C. (2003). Folkscience: Coarse interpretations of a complex reality. Trends in
Cognitive Sciences, 7, 368-373.
Keil, F. C. (2006). Explanation and understanding. In Annual Review of Psychology (Vol. 57,
pp. 227-254).
Laukkanen, M. (1994). Comparative cause mapping of organizational cognitions. Special
Issue: Cognition within and between organizations: Five key questions. Organization
Science, 5(Aug), 322-343.
Laukkanen, M. (1996a). Comparative Cause mapping of organizational cognitions.
Organization science, 5(3), 322-343.
Laukkanen, M. (1996b). Conducting Cause Mapping Research: Opportunities and challenges.
In B.Wintersheid & C.Eden (Eds.), Managerial and Organizational Cognition: new
directions in theory, methods and research. London: Sage.
Leiser, D. (2001). Scattered naive theories: why the human mind is isomorphic to the internet
web. New Ideas in Psychology, 19(3), 175-202.
Leiser, D., & Drori, S. (2005). Naive understanding of inflation. Journal of Socio-Economics,
34(2), 179-198.
Lewis-Beck, M. S. (1988). Economics and Elections: the major western democracies:
University of Michigan Press.
Lewis-Beck, M. S., & Rice, T. W. (1992). Forecasting Elections: CQ Press Washington, DC.
Moscovici, S. (1981). On social representations. In J. Forgas (Ed.), Social Cognition.
Perspectives on Everyday Understanding (pp. 181–210). London: Academic Press.
Lay understanding of macro-economics. 19
Moscovici, S. (1984). The phenomenon of social representations. In R. M. Farr & S.
Moscovici (Eds.), Social representations (pp. 3-69). Cambridge: Cambridge
University Press.
Purkhardt, S. C., & Stockdale, J. E. (1993). Multidimensional scaling as a technique for the
exploration and description of a social representation. In G. M. Breakwell & D. V.
Canter (Eds.), Empirical approaches to social representations (pp. 272-297). Oxford,
England: Clarendon Press/Oxford University Press.
Rozenblit, L., & Keil, F. C. (2002). The misunderstood limits of folk science: an illusion of
explanatory depth. Cognitive Science: A Multidisciplinary Journal, 26(5), 521-562.
Sevón, G. (1984). Cognitive maps of past and future economic events. Acta Psychologica,
56, 71-79.
Sturgis, P., Allum, N., Smith, P., & Woods, A. (2005). The Measurement of Factual
Knowledge in Surveys. Under review.
van Bavel, R., & Gaskell, G. (2004). Narrative and systemic modes of economic thinking.
Culture & Psychology, 10(4), 417-439.
Vergès, P. (1989). Représentations sociales de l'économie: une forme de connaissance. In D.
Jodelet (Ed.), Les représentations sociales (pp. 407-428). Paris: PUF.
Walstad, W. B., & Rebeck, K. (2002). Assessing the economic knowledge and economic
opinions of adults. The Quarterly Review of Economics and Finance, 42(5), 921-935.
Lay understanding of macro-economics. 20
Figure 1 - Causal map of one participant
Lay understanding of macro-economics. 21
V1
V10
V11
V12
V13
V14
V15
V16
V17
V18
V19
V2V3
V4
V5
V6
V7
V8
V9
Figure 2 - MDS Naives (Alienation = .32)
Lay understanding of macro-economics. 22
V1
V10
V11
V12
V13
V14
V15
V16
V17
V18
V19
V2
V3
V4
V5
V6
V7
V8
V9
Figure 3- MDS economists (Alienation = .197 )