For the knowledgeable reader it is always exciting to pick up a book by Douglas Hibbs. His new book deals with a subject of great interest to all Scandinavian political scientists and perhaps even economists: what is the micro basis for economic voting in Scandinavia?

However, in spite of its important subject, it is a strange little booklet. It is possible that it will disappoint most readers with the exception of the authors of this review. Two-thirds of the book deals critically with some work we are doing on Denmark, most of which is not yet available in a final version.¹ It is always very useful to have ongoing work commented on by competent colleagues, so we do not want to appear ungrateful. Also, Hibbs certainly makes a number of good points. Nevertheless, we are convinced that the main message of his book is empirically wrong.

It is also a strange book in another way. It reflects Hibbs's strong beliefs in the welfare state, and in the solidarity of all Scandinavians toward their fellow men - beliefs that have worn thin with many other observers. These beliefs are appealing from a moral point of view; but, as we shall show, they are not necessarily the most fruitful ones when it comes to answering empirical questions.

The book deals with the sociotropic/egotropic controversy in the micro theory of economic voting. Voting is egotropic to the extent people consult their pocketbooks. It is sociotropic to the extent it is based on people's perception of the economic well-being of society.

Characteristically, Hibbs prefers to speak of solidarity and egoism instead of the sociotropic and egotropic. He grudgingly accepts the term sociotropic in most of the book, but for egotropic he uses egocentric. He might as well have used the terms good and bad.² This is one indication of his value bias mentioned. It is also a controversial point: sociotropic behaviour may, but need not, be an expression of altruism or solidarity. The pioneering work of Kinder & Kiewiet (1981) already made this point, as Hibbs is well aware of.

The classical empirical result on economic voting (Kinder & Kiewiet 1979,1981) is that Americans are only sociotropic. A weaker version is that economic voting is predominantly sociotropic. This has been reported by Markus (1988) and by Lewis (1988), who covers the US, UK, Germany and France- Our central empirical result is that Danish economic voting is predominantly egotropic. In that respect Danes seem to be: different from most others, with the possible exception of Norwegians, as pointed out by Hibbs (pp. 46-49). None of the other studies covers a full-fledged welfare state of the Nordic type (Esping-Andersen 1991). We suggest that our results may generalize to other welfare states of this type.

Hibbs's main message is that Danes must be sociotropic. His book is one long

attempt to prove that point. To him the key thing is that Denmark is a welfare
state, and "it is difficult not to interpret the Danish welfare state as quite dramatic
evidence of the politics of solidarity" (p. 64).3

Most of the arguments offered by Hibbs share one common trait: they are conjectural. Hibbs has frequent hunches that if we had done this or that, then we would have got the result he knows, in his guts, to be right. This is a further peculiarity of the book: Hibbs consistently tries to make a substantive, empirical point about the nature of Danish economic voting by recourse to theoretical and methodological reasoning only. Thus, there is continuous theoretical and methodological barking. As we shall show, there is no real empirical bite.

The main thrust of Hibbs's assault is directed against two points: (1) the research design, and (2) the measurement of sociotropic concerns. These two lines of attack are not surprising. They aim at the two weakest points in most research of economic voting. They are always worth discussing.

Kramer vs. Hibbs. The Case of the Research Design

In the two papers discussed by Hibbs (NPI; NP2), we use a purely cross-sectional design. This design underlies the seminal work by Kinder & Kiewiet (1979, 1981). Up until now it has, by far, been the most common design in micro studies of economic voting. It is, for example, used in the widely quoted comparative study by Lewis-Beck (1988). Anybody, who wants to reach comparative results, must inevitably use the Kinder & Kiewiet model. We all know that this model is problematic, for reasons to be discussed, but it is the benchmark. Hibbs uses it too, but in his own way: he accepts, with minor qualifications, all evidence from crosssectional studies showing that economic voting is sociotropic. At the same time he rejects our finding, that Danes are egotropic, because they are reached using a cross-sectional design! We shall name this the Hibbsian double standard. It cannot but reinforce the suspicion of a value bias that was suggested by his choice of terminology already.

The use of a purely cross-sectional design in micro studies of economic voting was first criticized by Kramer (1983). It is not surprising that Hibbs marshals the authority of Kramer against our result. Hibbs presents Kramer's critique very elegantly. The two most relevant points are:

(i) The egotropic and sociotropic effects are something happening over time. Crosssections are timeless. Therefore, they cannot yield valid estimates of these effects. One has to use models with an explicit time index, and mixed timeseries/cross-sections data (pooled cross-sectional design).

(ii) Estimates of egotropic and sociotropic effects have several likely biases. These
biases may make the egotropic effect appear too small and the sociotropic too
large. So, perhaps, the Kinder & Kiewiet result is wrong.

One might argue that the well-known voter's myopia should make the problem (i)
small, but it is, in principle, a serious one that has to be carefully investigated.
There are two reasons for the biases (ii):

(a) The egotropic effect might drown in noise. Noise, in Kramer's sense, is all
"politically irrelevant" factors.

(b) In contrast, the sociotropic effect might be overestimated, because, as Markus (1988, 140) aptly paraphrases the argument: "much of the cross-sectional variation in beliefs about the state of the national economy is likely to be either spuriously related to the vote (i.e., influenced by long-held partisan identifications) or else attributable to error arising from the inevitably imprecise measurement of attitudes".⁴

The Kramer critique of our studies (NPI; NP2) thus implies that we underestimate the extent to which the egotropic effects dominate the sociotropic ones. Obviously, this is the opposite of what Hibbs wants to argue. Hence, he has to conjecture that in our case it must be the other way round. So he does.

We need not go into the details of Hibbs's endeavours to make his case stick by plausible conjecturing. The acid test of the validity of his guess must be the empirical evidence, as stated by Hibbs himself (p. 53). As it happens, we have provided such evidence. Actually, it is in the only piece published so far (NP4). It appears in Hibbs's list of references, but nowhere else in his text. In this study we have used the alternative design recommended by Hibbs (among others), namely a pooled cross-sectional design.

With the pooled cross-sectional design we obtain results that in all relevant respects are extremely close to the results found in the cross-sectional studies. Again there is a very strong egotropic effect, but only a weak sociotropic one. The egotropic effect found in the pooled cross-sectional study is even a little stronger than the one that emerges from the purely cross-sectional studies. This is, of course, what one would have predicted from Kramer's original argument. It is not according to Hibbs's hunches.

We think it remarkable, and very gratifying, that the cross-sectional and the pooled cross-sectional studies return results which are as close as they turn out to be. It is a potentially important finding and it tallies with our knowledge about voters' myopia. It also tells us that it takes more than just a change of research design to bring out that Danes do vote predominantly from sociotropic economic concerns.

Measuring the Sociotropic Variables: Are they Elusive or Robust?

The second recurring issue in Hibbs's critique deals with the measurement of
sociotropic economic concerns. He raises two major points.

The first point is that in two of our studies he considers the sociotropic variable is allegedly not well measured. Thus the dominating effect of egotropic variables found in our studies might be an artefact attributable to the way we measure the sociotropic concerns.

We readily admit that trying to measure respondents' sociotropic economic concerns is a frustrating experience. This is primarily because the very concept is seriously under-theorized. Therefore, it may be operationalized in (too) many different ways. Whatever operationalization you choose, it can always be argued that it is not the right one. It is of little help to turn to the literature in the hope of finding operationalizations of sociotropic economic concerns that can be seen to have worked for others. Since most of the literature consists of cross-sectional studies, you cannot, according to the Kramer-critique, trust any of the findings

reported there. Hence, unless you resort to the Hibbsian double standard, in most cases you cannot tell if a particular operationalization has really worked. Thus there is no "correct" way to measure sociotropic economic concerns. Neither is there a generally accepted and valid practical standard.

Under these conditions the basic question must become one of robustness. Is the main result empirically robust across a range of possible and plausible operationalizations, or does it depend upon a particular way of measuring the sociotropic effect? In the studies listed we work with four different operationalizations of sociotropic economic concerns. As far as the main result is concerned, we obtain the same result whatever the operationalization tried: with Danish respondents, the egotropic effect dominates the sociotropic one.

Furthermore, in the most recent of our interview rounds (May 1993) we included yet another variant of an old question. It is very close to the one that has been commonly used in American surveys to tap sociotropic motivations (and in the Eurobarometer as well).⁵ When controlled for party identification, responses to this question were found not to be significantly related to vote intentions. This brings our score up to five different measures tried - all with the same result. We have to admit that we have not tried everything. That is impossible in a finite lifetime. However, we have done quite a deal, and after a while it gets tedious to obtain the same result over and over again.

A second point repeatedly raised by Hibbs (pp. 23, 37, 49) concerns the proper interpretation of the government competency question. In (NP2) we include a variable that measures how competent people think the government is. We find that competency is an important intervening variable between economic evaluations and vote intentions. Hibbs claims that our government competency question really measures the respondents' sociotropic concerns. And, as he adds triumphantly, "By this interpretation, Nannestad & Paldam (1992) demonstrate right off the bat, so to speak, that sociotropic orientations largely govern the Bloc Voting behavior of Danes" (p. 23).

We beg to disagree. Hibbs's interpretation is based on the similarity between the wording of our competency item and of one of the items used by, inter alios, Kinder & Kiewiet to measure sociotropic motivations. However, there is one difference, which Hibbs overlooks. The American question refers to competency with respect to specified economic problems, like unemployment and inflation. In an American setting this will probably be understood as referring to social rather than personal concerns. In contrast, there is no reference to a specified problem in our wording of the item. Thus, with our question it is entirely left to the individual respondent to choose the factors by which to judge the government's competence. They may be sociotropic or egotropic - that is something to be tested empirically, as we do. The results show that among economic concerns personal ones dominate competency judgements.

Hibbs's misinterpretation of the item shows that he is neglecting a crucial difference between the USA and the Nordic welfare state: A key characteristic of a Nordic-type welfare state is that the State has taken responsibility for the situation of each individual. We think that people recognize that. Therefore, they might interpret the government's competency in either egotropic or sociotropic terms.

In summary, the weak impact of sociotropic motivations on vote intention and government trust is a robust finding across five different operationalizations. Hibbs's additional claim that the government competency question must be measuring sociotropic concerns is unsubstantiated.

Value for Money?

As stated at the very beginning of this review, other readers than us may feel that, because to its special character, Hibbs' book does not give them their money's worth. We have to concede that major parts of it will hardly be of interest to anyone other than him and us. With respect to the central empirical question whether the attitudes of the Danes towards government (and opposition) are determined predominantly by egotropic or sociotropic concerns, Hibbs does not make good his claim that sociotropic concerns must be the really important ones. We would like, nevertheless, to point to two aspects of the book that could be of more general relevance.

In the first place, Hibbs's critique of the main research design in the literature highlights some sore spots in the current research practices in the field of the economics of voting. Although, for the case at hand, empirical evidence contradicts his convictions and hunches, the points he raises remain important in a more general sense. One may use this book as a checklist telling in what respects one must make sure that results are: robust before going public. Perhaps Hibbs's hunches should be interpreted as nothing more than caveats. Then they could be extremely useful.

Secondly, there is also a lesson to be learned from the Hibbsian double standard. Hibbs's book here communicates an important message: it vividly illustrates that it is problematic to conjecture from a strongly held point of view. There is always a danger that the rules of scientific discourse come to give way to the logic of Christian Morgenstern's poetic figure Palmström:

Und er kommt zu dem Ergebnis Nur ein Traum war das Erlebnis Weil, so schliesst er messerscharf, Nicht sein kann, was nicht sein darf.⁶

---

NOTES

1. Our papers in the project exist in various versions. We hence refer to the papers that have been available to Hibbs, while he wrote his book, as (NPi), where i = 1, . . . , 4. We are now working on three more papers in the project, so Hibbs's book is very useful to us.

2. The strong moral connotations that easily creep into the terms make it problematic to ask people directly. It was tried by Observa 6/4-1992, as published in BT 3/5-1992, and a great majority answered that they voted according to solidarity and not egoism. People are certainly good and not bad, if you ask them directly. Therefore, as far as possible, let us keep moral judgement out of the terms.

3. It is, as already mentioned, a controversial point. Other authors such as Esping (1991) and Offe (1984) have not found it difficult to explain welfare states by other factors than solidarity. To get around these discussions, we prefer the terms egotropic and sociotropic.

4. In our discussion of Kramer's critique (NP4) we argue that (b) is much stronger than (a). It is very likely that people's perceptions of the whole economy are influenced by their personal experiences. If this is the case, the sociotropic effect snatches some of the egotropic coefficient. A perception bias hence means that the estimate of the sociotropic effect becomes too big, whereas the estimate of the egotropic effect becomes too small.

5. The wording was "How do you think the general economic situation in this country has changed since the new government took office?" The only difference to the American version is the time reference which used to be 12 months in American surveys.

---

6. Regrettably, we do not know of any English translation. Without artistic pretension, the meaning is: And he comes to the result/that his experience had been a dream/ because, as he infers with great acumen,/what should not be, cannot be.

REFERENCES

Esping-Andersen, G. 1991. The Three Worlds ofWelfare Capitalism. Oxford: Polity Press.

Kinder, D. R. & Kiewiet, D. R. 1979. 'Economic Discontent and Political Behavior: The
Role of Personal Grievances and Collective Economic Judgements in Congressional
Voting', American Journal of Political Science 23, 495-527.

Kinder, D. R. & Kiewiet, D. R. 1981. 'Sociotropic Politics: The American Case', British
Journal of Political Science 11, 129-161.

Kramer, G. H. 1983. 'The Ecological Fallacy Revisited: Aggregate- versus Individual-level
Findings on Economics and Elections, and Sociotropic Voting', American Political Science
Reviewll, 92-111.

Lewis-Beck, M. S. 1988. Economics and Elections: The Major Western Democracies. Ann
Arbor: Michigan University Press.

Markus, G. B. 1988. 'The Impact of Personal and National Economic Conditions on the
Presidential Vote: A Pooled Cross-Sectional Analysis', American Journal of Political
Science 32, 137-154.

Nannestad, P. & Paldam, M. 'Economic Grievances and Government Support in a Small
Welfare State: Micro Theory and the Danish Case.' 1991. (NPI).

Nannestad, P. & Paldam, M. Mellem samfundssind og egeninteresse: Samfundsøkonomi, privatøkonomi og mistillid til regeringens kompetence. Third version. To be published in Økonomi & Politik, 1993. Part of the Rockwool Foundation study of Political Distrust. (NP2).

Nannestad, P. & Paldam, M. 'The VP-Function. A Survey of the Literature on Vote and
Popularity Functions after 25 Years'. Forthcoming in Public Choice, proofread. (NP3).

Nannestad, P. & Paldam, M. Sammenhængen mellem samfundsøkonomi, privatøkonomi
og tilslutningen til regeringspartierne i Danmark 1990-91 - og nogle problemer i forbindelse
med at måle den korrekt. Politica, 25 årg, no 2, 189-206, Maj 1993. (NP4).

Offe, C. 1984. Contradictions of the Welfare State. London: Hutchinson.

Peter Nannestad, University of Aarhus

Martin Paldam, University of Aarhus