Age–price profiles for Canadian painters at auction

Abstract

We conduct an empirical analysis of the effect on the auction price of a Canadian painting of the age of the painter at the time of creation of the painting. We consider over two hundred artists, active over the entire history of Canadian art, who are pooled in the estimation of a hedonic regression in which a polynomial function in age enters as a regressor along with several other control variables. We consider the possibility that the age–price relationship has changed over time by: (a) estimating separate age–price functions for three generational groups of artists—those born before 1880, between 1880 and 1920, and after 1920 and thus coming of age in the world of post-war ‘contemporary art’ and (b) estimating a parameterization where the shape of the age–price profile is permitted to change continuously depending on the year of birth of the artist. Our principal result is that artists born more recently tend to ‘peak’ earlier in their careers than those of previous generations.

1 Introduction

It has been widely observed that the quality and importance of the work of a given artist can depend on the age of the artist when the work was executed. The pattern linking age to quality can of course depend on idiosyncrasies of the individual artist, but there may be common elements affecting many artists in the same way. The historical and national schools to which they belong, educational and market standards, membership in specific artistic movements, and the stylistic character of the artist’s work can all potentially exert systematic influences on the career pattern of the quality of work.

Czujack (1997) showed that age effects could be important in predicting prices in the context of a hedonic regression of auction prices for the paintings of Picasso. Broader groups of artists are studied in Galenson (2000, 2001) and Galenson and Weinberg (2000, 2001). The principal focus of this work is the possible existence of, and explanations for, changes in age–price profiles for different cohorts whose careers overlap during periods of particular art historical importance. Two examples of such periods are studied—nineteenth- and early twentieth-century Paris and post-war New York City. These studies find that, in both historical periods, earlier generations are characterized by an age–price profile well to the right of that of younger generations, that is to say, that the members of earlier generations tend, on average, to ‘peak’ later in their careers than do younger painters. These econometric results are obtained from the estimation of hedonic regressions, which include polynomials in painters’ ages.

The interpretation given by the authors to these results (see also Galenson (2006, 2009)) is that there was a shift in both Paris and New York in the mode of artistic creation, from work whose innovation is ‘experimental’ in nature, to work that is ‘conceptual’. The former category includes art works where the material execution of the work is of paramount importance: handling of paint, pictorial and coloristic effect, and spontaneity of execution. Conceptual art, on the other hand, is distinguished by the originality and interest of the idea or thinking, which is embodied in the work. Conceptual artists carefully plan and think through each major work, of which there will often be a relatively small number, with the execution of the work being a more or less mechanical application of the already developed ideas. Artists are seen as having personal pre-dispositions to one or the other of these tendencies, with artists with a more conceptual orientation being more likely to do their best work at a relatively younger age. The main reason is that experimental art relies on the development of a mastery of the technical means of applying paint and working with colour, whereas a conceptual artist solves conceptual problems in clever and original ways, a talent associated with young people unencumbered by ingrained patterns of thought.

This model of categories has attracted much attention and has had many critics. Ginsburgh and Weyers (2006) note that the distinction between conceptual and experimental artists seems similar to or parallel to that between disegno and colore in Italian Renaissance aesthetics and the seventeenth-century distinction between ‘drawing’ and ‘colour’. We can also refer to the distinction made in the 1920s by Wölfflin (1950) between ‘classical’ and ‘baroque’ art. Like Galenson, Wölfflin identifies a shift in the general mode of aesthetic creation, in this case from the sixteenth century, whose art is characterized as classical (with associated formal properties of linearity, planarity, closed form, multiplicity, and clearness), to the seventeenth, whose baroque art possesses the opposite qualities of painterliness, recession, open form, unity, and unclearness.

To the extent that Galenson’s conceptual/experimental dichotomy can be mapped onto the aesthetic dichotomies mentioned above, he is claiming that stylistic shifts occurred in Paris and New York that go in a direction opposite to the shift described by Wölfflin around the year 1600 for Europe as a whole. In this regard, it is of interest to consider a lecture delivered by Clement Greenberg in 1968 on the avant-garde art of the 1960s:

…advanced art of the 60s subscribes almost unanimously to these canons of style… that Wölfflin would call linear… Think by contrast of the … 1950s: the fluid design or layout, the ‘soft’ drawing, the irregular or indistinct shapes or areas, the uneven textures, the turbid color. (Greenberg 1993, p. 294)

It is therefore plausible that general shifts in artistic style, coincident with the shift in age–price profiles found by Galenson, occurred in Paris and New York. The question is whether the shift from painterly to classical styles is necessarily connected with more youthful artists. In both cases, the shifts were associated with the emergence of an ‘avant-garde’ mentality of artistic creation. As analysed by Jensen (1994), the ‘avant-garde’ is an inherently youthful philosophy that revolts against currently prevailing artistic styles and which has as a necessary condition the presence in a large city of a large number of young artists (a ‘cluster’—see Hellmanzik (2010a)). The avant-garde emerged in Paris at a time when an impressionistic aesthetic was predominant, so that a consciously revolutionary youth movement would naturally adopt an opposing aesthetic—a classicist one. The idea of the avant-garde took hold in New York in the early post-war era when the leading artists were abstract expressionists, so that a reaction such as that described by Greenberg would also be natural.

We may then ask whether the avant-garde mentality necessarily corresponds to the production of art which is of a ‘conceptual’, or ‘linear’, aesthetic. In the two cases analysed by Galenson, it could just be a historical coincidence that the avant-garde emerged during a period of a predominantly painterly aesthetic. May it not be possible that a subsequent generation of young artists would choose to react in its turn by a return to painterliness and have we not perhaps seen this in the resurrection of painting after it had been declared dead by the 1960s–1970s generation of Conceptualists?

We will maintain below, in our analysis of Canadian artists, that a young and dynamic avant-garde need not be inconsistent with a painterly, experimental approach. We do so within a broader historical perspective of the evolution of artistic styles than has heretofore been considered. The studies mentioned above [as well as Accominotti (2009) and Hellmanzik (2010b)] only examine modern artists working in large cities that are already established as clusters. Furthermore, the changes in career profiles of artists found by Galenson have the property of one-time discrete shifts. However, it may be of interest to see whether these shifts are actually just phases in a longer historical process, which would include the rise to cluster status of the places considered.

An analysis of age–price profiles for Canadian artists over time may be of relevance here. The demographic, economic, and artistic evolution in Canada over the century beginning in 1850 was very rapid. From a sparsely populated colonial society, with a limited art market, it steadily grew in population, urbanization (see Table 1), and scale of its artistic community to the point where by 1960 it had a number of avant-garde artistic clusters. Changes in the conditions of artistic production had important consequences for the careers of artists. Our principle conclusion will be that there is a continual shifting to the left of the age–price profile for Canadian artists over a period of over a century (which we will measure through a hedonic regression specification in which artists are partitioned into three discrete historical ‘cohorts’, as well as a specification in which the age–price relationship shifts continuously as a function of birth date), but that there is no such continuous shift in aesthetic styles towards progressively more conceptual production. In fact, we will see that the most painterly, ‘experimental’ art ever produced in Canada was due to a very young group of avant-garde painters. This suggests that the hypothesis drawn from the observation of pre-war Paris and post-war New York, viz., that young avant-gardes are necessarily associated with a classicist, linear, or conceptual aesthetic, may not possess universal validity.

Table 1 Total and urban population of Canada, 1851–1971

2 Data

Records of sales of Canadian paintings at auction from 1968 to 2010 were collected from Campbell (1970–1975, 1980), Sotheby’s (1975, 1980) and Westbridge (1981–2011). Our data set includes results on sales for painters judged to be of significant interest from the standpoint of Canadian art history, this criterion being satisfied if a painter is mentioned in one of the major histories of Canadian art written by Harper (1977) or Reid (1973, 1988) and the overviews of contemporary painters in Murray (1987) and Nasgaard (2007). We consider only oil and acrylic paintings and only sales for which the auction house provides a secure attribution. For each painting, we recorded, in addition to the identity of the artist, the height and width, the medium and support, the auction house, the date of sale, the genre of the picture, and, when available, the date of execution of the painting. The prices we use are hammer prices as reported in the aforementioned publications. The resulting data set, an expanded version of that used in the study of the investment properties of Canadian paintings by Hodgson and Vorkink (2004), contains 26,955 sales. After taking all dated paintings, and excluding every artist for whom less than five dated sales are available, we are left with a sample size of 10,568.¹ These latter form the sample on which the hedonic regressions reported below are based. There are 211 painters for whom at least five dated paintings are recorded, and they are listed in the ‘Appendix’ in alphabetical order along with birth date and number of observations.

The data are summarized in Table 2, where we can see that the average age of the painter at time of execution for the 10,568 sold paintings in our sample is just above 50, with a standard deviation of 16. The age distribution of paintings that are sold at auction differs significantly for the most recent generation, born after 1920 (mean of 42.7 and standard deviation of 11.4), compared with the earlier ones (respective means of 48.9 and 53.7 and standard deviations of 15.2 and 16.4 for artists born before 1880 and between 1880 and 1920). Although this result could be due in part to bias arising from the fact that some members of this latest cohort were alive and active for all or part of our sample period, it is robust to the joint truncation of the set of artists included to those born before 1935 and of the dates of sale to those after 1995. It suggests a number of hypotheses consistent with the notion that the generation born after 1920 is more likely to do its best work at a relatively early stage of life: (i) a decline in quality, originality, or financial remuneration of an artist’s work will tend to lead to a decline in productivity measured by number of works painted; (ii) an early period of frenetic productivity could lead to burnout; (iii) later works, if of lesser quality or historic interest, will be less demanded in the secondary (auction) market.

Table 2 Average age at time of execution for paintings sold at auction, by cohort

3 Historical overview of Canadian artists

In this section, we attempt to characterize the principal stylistic currents in the history of Canadian art and their linkages, if any, with career profiles of artists. From the late colonial period through the Confederation of 1867 and up to World War I, Canada saw rapid demographic, territorial, urban, and economic expansion. There was an associated increase in the demand for art and in the number of artists. The demand was principally for landscapes [see Brooke (1989) and Reid (1979b)], which is where stylistic trends in Canadian art can best be traced.

Few Canadian artists of this period could make a living solely from the sale of art, although as the cities developed there was a steady increase in job opportunities in the fields of art teaching and commercial illustration. Most Canadian artists of note were educated in European academies, principally in Paris, and would spend as much time in Europe as was economically feasible. These artists would develop towards their peak relatively slowly. Upon return to Canada, one needed to work for a living while trying to find time to paint. The art market was thin, with few galleries and with art fairs and exhibition societies as the principal means of finding buyers. These problems became steadily less acute as cities and incomes grew. Thus, we would expect, for simple reasons of market development, that artists would have more opportunity earlier in their lives to produce better work as time passed.

However, there is no evident stylistic shift during the pre-World War I period from a more baroque, or experimental, to a more classical or conceptual aesthetic. The earliest group of painters was fairly diverse stylistically. For example, Cornelius Krieghoff had an extremely varied career before finally finding a market niche working in a painterly, baroque style for collectors in colonial Quebec (see Harper (1979) and Reid (1999)). In the context of a very thin art market, he had to patiently develop a personal style and find a client base, and did his best work in his early- to mid-40s.

We may classify Krieghoff as ‘experimental’, but the principal reason that he peaked so late in life is the economic environment in which he worked. One could of course counter that it was an environment propitious to experimental artists and that Krieghoff’s success was due to a process of self-selection. However, if we consider an artist such as Lucius O’Brien, whose painting Sunrise on the Saguenay (National Gallery of Canada) is one of the most striking masterpieces of nineteenth-century Canadian art, we observe a conceptual inclination. O’Brien was born in rural Ontario and only decided at age 40 to pursue a career in art. His masterpiece was produced 8 years later, very early in his career as an artist, and thus consistent with our characterization of him as a conceptual artist. This will not appear in an age–price profile; however, because of how late in life O’Brien decided to turn to art. Indeed, artists of this period had a variety of career profiles, educational backgrounds, and stylistic orientations. In an immature and thin art market, where no standards have yet been established regarding the artistic profession, we will tend to observe profiles weighted to the right and this regardless of aesthetic orientation.

The post-confederation, ‘French’ stage in Canadian art (Reid 1988) was characterized by the methods taught in the Parisian academies, those of the European juste milieu (Jensen 1994). This ‘Canadian Impressionism’ (Duval 1990) can be labelled ‘experimental’, due to the impressionist aesthetic, and in that the artists sought to adapt impressionist brushwork to the depiction of Canadian subjects. The most successful were Maurice Cullen, James Morrice, and Marc-Aurèle de Foy Suzor-Coté [see Antoniou (1982), Cloutier (1985), Hill (1992), and Lacroix (2002)]. We would expect such artists to do their best work at a relatively advanced age, so that the majority of artists active before World War I (and born before about 1880) should peak relatively late, and we group these artists as a single ‘cohort’.

Note that the effects of isolation and a thin market could still be important for an artist born as late as 1871. Emily Carr lived in Victoria, British Columbia, a small city with no art market. She abandoned painting for several years, only resuming major artistic activity after the age of 50, with the work she created in her fifties and sixties generally considered to be her best (Thom et al. 2006; Shadbolt 1979).

By 1910, Montreal and Toronto had reached a size where a critical mass of young artists could find employment in such professions as commercial design or teaching, allowing them to form groups that could work together, exhibit and discuss their work, share studio space, and generally benefit from being present in a nascent artistic ‘cluster’. In this environment, the first true movement (in the narrow sense of a cohesive group of artists working closely together on a coherent and well-specified aesthetic programme) in Canadian art history emerged in Toronto. The group of eight painters (Lawren Harris, J. E. H. MacDonald, Frank Johnston, A. Y. Jackson, Arthur Lismer, Fred Varley, Tom Thomson, and Frank Carmichael) had formed by 1914, but, due to the interruption of the war and the death of Thomson, had their first formal joint exposition in 1920, calling themselves the Group of Seven [see Hill (1995) and King (2010)]. Their formal approach was based on stylized linear patterning and bright colours, similar to Scandinavian interpretations of post-impressionism (Nasgaard 1984).

The group’s most original member was Tom Thomson, who only took up painting at the age of 35 (see Reid and Hill 2002). In five years, he made rapid progress towards a rigorously planar and linear compositional style. His paintings the Jack Pine (1917, National Gallery of Canada) and the West Wind (1917, Art Gallery of Ontario) are iconic masterpieces that allow us to place him in the ‘conceptual’ category, with a classical aesthetic and highly composed masterpieces at a very early stage of his career. An estimated age–price profile, however, would have him peaking at a relatively late age, as he emerged in an artistic environment that was in the process of arriving at a mature stage.

Many of the top Canadian artists of the interwar years opted for a formal approach that was more classical in nature, with cubist influences and geometric abstraction making their first appearances. One could name, for example, Charles Comfort, Carl Schaefer, Edwin Holgate, Alfred Pellan, Marion Scott, Bertram Brooker, and Lemoine Fitzgerald [see Hill (1975), Gray (1976, 1977), Reid (1979a), and Pepall and Foss (2005)]. It was now possible for artists to develop earlier in their careers, due to the presence of a critical mass of young artists, the ability to get training in Canada (or the United States), and to earn an income in an art-related occupation. There was also an expanding market for contemporary art among private collectors and expanding public museums. Based on these considerations, we would expect estimated age–price profiles for the post-1880 cohort to be to the left of those of the earlier one. It does seem that this shift, if indeed present, was accompanied by a move to a more classical or conceptual aesthetic, possibly due to the influence of international aesthetic trends in modern art.

However, some top interwar artists worked in a more painterly manner, for example John Lyman, Goodridge Roberts, and Jean-Paul Lemieux [see Borcoman (1969) and Grandbois (2001)]. The most important was Paul-Émile Borduas [see Gagnon (1976)], who around 1940 (at age 35) became very interested in the theory of automatist creation and developed an automatist method of abstract painting. He had a number of talented pupils, who adopted his approach with great success. By 1942, the Automatiste group, the second major movement in Canadian art history, had been established. It included Fernand Leduc, Jean-Paul Riopelle, Marcelle Ferron, Pierre Gauvreau, Marcel Barbeau, Jean-Paul Mousseau, and Françoise Sullivan [see Gagnon (1998) and Nasgaard and Ellenwood (2009)].

The Automatistes were mostly very young, and their method attempted to find the inner wellspring of the artist’s individuality, with as little control as possible from conceptual rationality. It was thus as far as could possibly be from any form of conceptual art. In practice, it resulted in a highly gestural and painterly coloristic abstraction that developed independently of American Abstract Impressionism. Of particular importance was its youthful nature—the spontaneity and independence from rational control required were to be most likely found in young people. The high-water mark of the group is thought to have been around 1948, a ‘heroic’ period when the median age of its members was 24.

The avant-garde arrived in Toronto a few years later with the founding of Painters Eleven [see Nowell (2010)], a third important movement but one that was less stylistically cohesive or ideologically motivated than the Group of Seven or the Automatistes. Its members (Alexandra Luke, J. W. G. MacDonald, Harold Town, Tom Hodgson, Kazuo Nakamura, Hortense Gordon, William Ronald, Jack Bush, Oscar Cahen, Ray Mead, Walter Yarwood) were diverse in age and background and worked in a very painterly and gestural style, and many obtained commercial gallery representation. Indeed, by 1960, an avat-garde commercial dealer system was also forming in the major Canadian cities. Some artists could now make a living selling their art, and further income opportunities existed due to new Canada Council grants and the expansion of the university system.

Hellmanzik (2010a) finds that cities with larger artistic communities—‘clusters’—are associated with artists producing better work earlier in their careers. Accominotti (2009) also stresses the importance of belonging to movements in the development of careers, with artists seeming to do their best work when the creative stimulation of working closely with other artists with similar interests is available. None of this is necessarily at odds with the Galenson thesis that younger artists tend to produce more conceptually oriented art: such a thesis would, in the context of the findings of Hellmanzik (2010a) and Accominotti (2009), indeed predict such a tendency in larger clusters or in movements of younger painters. In the case of Canada, where the development of a mature art market was a gradual process from the country’s colonial period up to 1960, one would predict a gradual shifting to the left of age–price profiles, an associated trend towards a more conceptual aesthetic, and more artistic movements composed of younger conceptual artists.

4 Econometric model

We estimate two different specifications of a hedonic regression model, in both of which log price is regressed on the various painting-specific characteristics listed in Sect. 2, with the fashion in which age enters the regression differing in the two cases. First, we use a fourth-order polynomial model, with the painters partitioned according to the following three broad cohorts or generations: (1) those born before 1880, (2) those born in the years 1880 to 1919, and (3) those born in or after 1920. Separate parameter estimates are computed for each cohort. The second model pools all the painters but now adds to the polynomial in age the same polynomial, multiplied by the date of birth of the painter. This has the effect of allowing the polynomial function linking age to price to change continuously through time (according to date of birth), rather than to change in two discrete jumps as in the first specification.

Both specifications can be written as special cases of the following general model:

$$ y_{i} = x_{i}^{\prime } \beta + g\left( {z_{i} ,\gamma } \right) + u_{i} , $$

(1)

for $i = 1, \ldots ,n$ where y _t is the log price for sale i, the number of sales is n (equal to 10,568 in our case), x _i is a vector of observations of k control variables (other than age), β is an unknown k-vector of parameters, z _i is the age of the painter at time of execution of painting i, g is a function whose form will be discussed in more detail below, γ is an unknown parameter vector, and u _i is a zero-mean disturbance, assumed to be independent of the regressors. The 306 regressors contained in the vector x include annual time period dummies, dummy variables for painter, auction house, medium and support, and genre, along with height, width, and surface area measures. Our concern in this paper is with the estimation and analysis of g, so we treat β as a nuisance parameter [see Hodgson and Vorkink (2004) for a more complete analysis of a hedonic regression similar to the one considered here].

In the first place, we estimate a fourth-order polynomial in age, where the parameters are permitted to differ for the three cohorts of painters described above. We therefore have

$$ g\left( {z_{i} ,\gamma } \right) = \sum\limits_{j = 1}^{3} {\left( {\gamma_{j1} z_{i} + \gamma_{j2} z_{i}^{2} + \gamma_{j3} z_{i}^{3} + \gamma_{j4} z_{i}^{4} } \right)I\left( {i \in {\text{gen}}(j)} \right)} , $$

(2)

where $I\left( {i \in {\text{gen}}(j)} \right)$ is an indicator function equal to one if the painter of painting i belongs to generation j, and equal to zero otherwise, and γ is the vector of the twelve parameters in Eq. 2. The quartic (fourth-order polynomial) specification is popular in empirical studies of age-earning or experience-earning profiles, known as Mincer equations after Mincer (1974), and Galenson and Weinberg (2000, 2001) estimate fourth-order polynomials in age for early twentieth-century French artists, as well as for post-war American ones. Hellmanzik (2010b) also uses a fourth-order model for age–price profiles of various groupings of modern artists.

The second specification is a quartic function in age, supplemented with date-of-birth effects:

$$ g\left( {z_{i} ,\gamma } \right) = \sum\limits_{j = 1}^{4} {\left( {\gamma_{j} z_{i}^{j} + \gamma_{j + 4} z_{i}^{j} {\text{db}}_{i} } \right)} $$

(3)

where db _i is the year of birth of the painter associated with observation i, and γ is the vector of the eight parameters in Eq. 3.² This specification conforms better to the discussion of the history of Canadian artists’ careers in the preceding section, where a gradual thickening of the market, or ‘clustering’, was associated with the possibility that career profiles would evolve gradually to the left rather than jumping discretely at certain historical turning points.

We have already referred to the study of Accominotti (2009), who maintains that an artist’s membership in a movement (defined in the narrow sense of a unified group of artists working closely together on a relatively coherent aesthetic programme) may have effects on his career dynamic beyond what would be predicted based on cohort-specific age effects. To check for this effect in our data, we also estimate versions of the two preceding models in which the regression is supplemented by separate quadratic functions of year of production for the members of the three most important movements in Canadian art history (viz., the Group of Seven plus Tom Thomson, the Automatistes, and Painters Eleven). The model in this case is:

$$ y_{i} = x_{i}^{\prime } \beta + g\left( {z_{i} ,\gamma } \right) + h\left( {v_{i} ,\delta } \right) + u_{i} , $$

(4)

where

$$ h\left( {v_{i} ,\delta } \right) = \sum\limits_{j = 1}^{3} {\left( {\delta_{j1} v_{i} + \delta_{j2} v_{i}^{2} } \right)I\left( {i \in {\text{mvt}}(j)} \right)} , $$

(5)

v _i is the year painting i was painted and $I\left( {i \in {\text{mvt}}(j)} \right)$ is an indicator function equal to one if the painter of painting i belongs to movement j, and equal to zero otherwise, and δ is the vector of the six parameters in Eq. 5, and where the model is estimated for each of the two specifications of g(z _i , γ) given in Eqs. 2 and 3.

Under all the above specifications, the regression is linear and can be consistently estimated by ordinary least squares (OLS). However, the errors in a model such as these are generally not identically distributed normal (preliminary OLS estimates of the model under specifications (Eqs. 2, 3) yielded Jarque and Bera (1980) normality statistics of 1648 and 1561, respectively), and so OLS will not be the maximum likelihood estimator and will be asymptotically inefficient. In particular, the presence of heteroskedasticity and non-normality in the errors suggests the use of an estimator that is robust to these departures from the canonical OLS assumptions. One could correct for the presence of heteroskedasticity in estimating the model by weighted least squares, which would be efficient if the weights were correctly estimated and the weighted errors normal. Both assumptions are, however, questionable. One could, for example, follow Galenson and Weinberg (2001) in weighting each observation by the standard deviation of realized prices for its painter, but this omits other possible sources of heteroskedasticity, such as the date of sale, as well as any non-normality that may be present. One could, in principle, deal with the first of these problems by weighting the observations by a non-parametric function of the regressors, but this introduces curse of dimensionality problems in the non-parametric estimation. One weighting scheme that will be considered, in order to address the possible distortion of results based on the substantial imbalance in the number of observations for different artists, is to downweight each observation by the square root of the number of observations for the relevant artist. We will report results of the above models with and without this weighting (except for Eqs. 4, 5, where estimates were not sensitive to weighting.)

We opt to estimate all versions of the regression using the semiparametric adaptive estimator developed by Bickel (1982). This estimator treats the errors as being independently and identically distributed (iid) from a distribution of unknown form and is fully asymptotically efficient under general conditions. However, Hodgson (2000) shows that Bickel’s (1982) estimator is robust to the presence of heteroskedasticity in the errors and will adapt for the non-normality induced in the unconditional density of the errors by this heteroskedasticity. Furthermore, the usual standard error estimates proposed by Bickel (1982) are also robust. Thus, the adaptive estimator, though not fully asymptotically efficient in the presence of heteroskedasticity, nevertheless has powerful robustness properties, can provide a substantial efficiency gain relative to OLS, and is easy to compute, as described in Hodgson and Vorkink (2004) [The estimator is also applied to hedonic regression by Hodgson et al. (2006)].

As an additional robustness check, we re-estimated each model in excluding all artists with less than 10 observations (as is done by Hellmanzik (2010b)), but the results were relatively insensitive in comparison with choice of weighting scheme and so are not reported.

5 Empirical results

The parameter estimates for the quartic age–price relation for the model estimated with artists partitioned by cohort are reported in Table 3 (for the regression estimated without weighting on observations) and Table 4 (with the weighting scheme mentioned in Sect. 4), and the associated estimated curves are plotted in Fig. 1a (no weighting) and b (with weighting). In the graphs, the profiles, which are identified up to vertical shifts, are normalized to equal zero at age 20. In all cases, the Wald statistic for joint significance of the age parameters is significant at the 1% level, as is the Wald statistic for each pairwise comparison of parameters across cohorts. This latter result implies that disaggregation by cohort is justified and that we get significantly different profiles by cohort. The finding that the peak age shifts to the left across cohorts is robust to estimation methods, the main differences being that, with weighting, the two earlier cohorts peak earlier and prices drop off more quickly past the peak. In both graphs, the pre-1880 cohort peaks later than, and prices subsequently drop less rapidly than, for the second cohort. For the pre-1880 cohort, the peak is 49 (without weighting) and 37 (with weighting), with respective peaks for the second cohort of 34 and 30. The third cohort peaks at 25 in both graphs.

Table 3 Hedonic regression: estimates of parameters of fourth-order polynomial in age, by cohort (unweighted observations)

Table 4 Hedonic regression: estimates of parameters of fourth-order polynomial in age, by cohort (weighted observations)

Fig. 1

Age–price profile by birth cohort. a Unweighted estimator, b weighted estimator

The results of the estimation of model (Eq. 3), in which date-of-birth effects are included, are reported in Table 5, and the age–price profiles are illustrated in Fig. 2a (no weighting) and b (with weighting) for three examples of years of birth, at 50-year intervals from 1820 to 1920. The date-of-birth effects are jointly significant at the 1% level for both estimators, although none of the eight parameter estimates is significant at even the 10% level for the weighted estimator (all of the unweighted estimates are significant at the 5% level and most also are at 1%). We can see from Fig. 2a and b a clear shift to the left in the peak age as function of date of birth, with the peak ages in Fig. 2a declining from 41 (for 1820 birth) to 36 (1870 birth) to 27 (1920 birth). The corresponding peaks in Fig. 2b are 38, 35, and 29. The major difference between the graphs is the rate of price decline past peak, which is estimated to be faster with the unweighted estimator.

Table 5 Hedonic regression: estimates of parameters of quartic in age for all artists with date-of-birth (D.B.) effects included

Fig. 2

Age–price profile with date of birth effects. a Unweighted estimator, b weighted estimator

The aforementioned results are all consistent with the basic hypothesis outlined in Sect. 3, viz. that profiles shifted to the left over time. The model with date-of-birth effects is probably more consistent with the idea that changes in career profiles were gradual, corresponding to gradual changes in demographic and economic factors underlying the art market. However, both of our cohort shift dates correspond to a period that begins with an artistic movement that was coherent and revolutionary in intent and may have helped provoke a more abrupt shift in style among younger painters.

The parameter estimates for the quadratic function in year of production for art works created by the three major movements (Eqs. 4, 5) are reported in Table 6 for the cohort model (Eq. 2) and in Table 7 for the model with date-of-birth effects (Eq. 3), with respective estimated curves plotted in Fig. 3a and b, normalized in both cases to peak at 2.0. In all cases, the quadratic coefficients are jointly significant at the 1% level, suggesting that for the members of these movements, there was a systematic effect on the career profile, due to membership in the movement, beyond what our aggregated age–price profiles would predict. The profiles are very similar in the two cases. The Automatiste peak in both cases is quite early, at or shortly after 1945, in the earliest and most revolutionary phase of the movement and when most of its members were in their early 1920s. The Painters Eleven group effect peaks in 1965, after the formal dissolution of the group in 1960. The Group of Seven peak is surprisingly late, in the 1935–1940 range. The Painters Eleven result could be explained in considering that the movement itself was the least coherent and ideological of those considered here, that one of its main goals was to try to develop a market for abstract art in Toronto, and that it was only after 1960 that such a market became fully developed. The Group of Seven result is harder to account for.

Table 6 Hedonic regression: estimates of parameters of quadratic in year of production for three artistic movements (regression with cohorts)

Table 7 Hedonic regression: estimates of parameters of quadratic in year of production for three artistic movements (regression with date of birth effects)

Fig. 3

Year–price prifile. a Three movements model with cohorts, b three movements model with year-of-birth effects

6 Conclusion

Recent work has shown that the environment in which an artist works can have important implications for the career pattern of the artist’s productivity. Hellmanzik (2010a, b) and Accominotti (2009) find that the presence of an artist in an active artistic centre, in which possibilities for interaction with other artists and market participants, allowing for the formation of groups and movements, will have systematic effects on the timing of the artist’s best work. It will generally come earlier in life and during periods of membership in creative movements. Combined with the hypothesis developed in various works of Galenson (2000, 2001, 2006, 2009) and Galenson and Weinberg (2000, 2001), according to which younger artists have a comparative advantage in the production of more conceptually oriented work [which would tend to have the aesthetic correlative of art that is more classical or linear in nature, in reference to long-standing aesthetic dichotomies described by Ginsburgh and Weyers (2006) and more recently associated with Wölfflin (1950)], we would expect a more classical or conceptual aesthetic to prevail in more highly developed artistic centres or ‘clusters’.

Most of the existing literature concerns already well-developed clusters that transition to avant-garde markets, following the transition in Paris around the year 1900 described by Jensen (1994). We extend the literature by considering a market that in a little over a century made a progressive transition from an artistically and culturally immature colonial society, to a country with large cities and maturing art centres during the interwar period, to an avant-garde market beginning around 1940 and fully developed and integrated into international markets and aesthetic trends by 1960. We find a steady and progressive shift to the left of artists’ age–price profiles during this period, consistent with the first hypothesis, regarding the effect of clusters.

Our conclusions regarding the effect of market development and the consequent changes in career profiles on the aesthetic character of artistic output are less clear-cut, not least due to the proviso that must be made with regard to our degree of confidence in any aesthetic judgment. Although it seems possible to detect a certain shift towards a more classical aesthetic after World War I, the first movements working in the avant-garde phase of the 1940s and 1950s were very painterly and expressionistic in nature. In the case of the Automatiste group, the major breakthroughs were made when most members of the group were extremely young, not much past 25. The artists’ youth was seen as an essential component of their success, as the spontaneity, authenticity, and radical anti-conceptualism of the artistic expression were held by the movement’s advocates to be hallmarks of youth. This consideration, if it does not stand on its head the Galenson hypothesis (i.e. that youth have a comparative advantage in conceptual art), at least suggests that that hypothesis is too simplistic, as it omits from consideration this category of expressionistic art.

The stylistic history of Canadian art after 1960 is similar to that elsewhere, as Canada had by this date become a participant in the contemporary avant-garde ‘art world’. The trend towards more geometric art in the 1960s detected by Greenberg (1993, p. 294), along with the rise in the late 1960s and early 1970s of conceptualism as a definite artistic movement, was also present in Canada, as illustrated by a comparison of the exhibition catalogues of two recent surveys of 1950s and 1960s art in Canada (Leclerc 1992; Leclerc and Dessureault 2005). There have been periodic returns of painterly painting, but it is too early to judge as to whether contemporary artists’ career profiles are dependent on the stylistic category of their art. If the historical importance of Picasso’s final work is still being sorted out, it could be some time yet before we can form conclusions about the careers of artists who have only just emerged in the last few decades.

Appendix: List of painters included

(the painters included in our empirical study are listed here in alphabetical order, with year of birth being provided in parentheses along with the number of observed sales for each painter)

Carl Ahrens (1863, 16), Edmund Alleyn (1931, 47), William Atkinson (1862, 81), Marcel Barbeau (1925, 31), J.M. Barnsly (1861, 27), Robert Bateman (1930, 9), Maxwell Bates (1906, 150), William Beatty (1869, 98), Henri Beau (1863, 23), Leon Bellefleur (1910, 132), Frederic Martlett Bell-Smith (1846, 54), Louis Belzile (1929, 15), Andre Bieler (1896, 53), B.C. Binning (1909, 26), David Blackwood (1941, 7), Ronald Bloore (1925, 10), Bruno Bobak (1923, 13), Molly Lamb Bobak (1922, 32), David Bolduc (1945, 15), Paul-Emile Borduas (1905, 60), Fritz Brandtner (1896, 16), Fred Brigden (1871, 6), Miller Brittain (1912, 7), Bertram Brooker (1888, 12), Archibald Browne (1862, 28), Franklin Brownell (1857, 56), William Blair Bruce (1859, 15), William Brymner (1855, 26), Dennis Burton (1933, 18), Jack Bush (1909, 75), Frank Carmichael (1890, 56), Emily Carr (1871, 97), A.J. Casson (1898, 499), Jack Chambers (1931, 5), W.H. Clapp (1879, 36), Paraskeva Clark (1898, 24), F.S. Coburn (1871, 267), Alex Colville (1920, 12), Charles Comfort (1900, 71), Stanley Cosgrove (1911, 259), Graham Coughtry (1931, 9), William Cresswell (1822, 20), Maurice Cullen (1866, 75), Jean Dallaire (1916, 61), Georges Delfosse (1869, 11), Rodolphe de Repentigny (1926, 9), Jacques de Tonnancour (1917, 51), Joseph Drapell (1940, 5), Rodolphe Duguay (1891, 24), Albert Dumouchel (1916, 11), Alma Duncan (1917, 7), Allan Edson (1846, 19), Patterson Ewen (1925, 20), Ivan Eyre (1935, 12), A.S. Falardeau (1822, 7), Gathie Falk (1928, 5), Harold Feist (1945, 6), Marcelle Ferron (1924, 86), Lemoine Fitzgerald (1890, 26), John C. Forbes (1846, 7), Tom Forrestall (1936, 12), J.W.L. Forster (1850, 8), Michael Forster (1943, 18), Marc-Aurèle Fortin (1888, 82), Paul Fournier (1939, 10), Joseph Franchere (1866, 10), John A. Fraser (1838, 6), Louise Gadbois (1896, 65), Clarence Gagnon (1881, 123), Yves Gaucher (1934, 8), Pierre Gauvreau (1922, 14), Charles Gill (1871, 10), Ted Godwin (1933, 10), Betty Goodwin (1923, 7), Hortense Gordon (1887, 16), Richard Gorman (1935, 15), Kate Graham (1913, 7), John Hall (1914, 7), John Hammond (1843, 137), Lawren S. Harris (1885, 163), Robert Harris (1849, 45), Doug Haynes (1936, 11), Adrien Hebert (1890, 32), Prudence Heward (1896, 6), Randolph Hewton (1888, 13), Tom Hodgson (1924, 16), Edwin Holgate (1892, 66), Yvonne McKague Housser (1898, 45), E.J. Hughes (1913, 109), Jack Humphrey (1901, 14), Charles Huot (1855, 10), Jacques Hurtubise (1939, 8), Gershon Iskowitz (1921, 36), A.Y. Jackson (1882, 858), Otto Jacobi (1812, 85), Alex Janvier (1935, 22), Donald Jarvis (1923, 7), C.W. Jeffreys (1869, 5), Jean-Paul Jerome (1928, 20), Frank Johnston (1888, 142), Illingworth Kerr (1905, 239), Dorothy Knowles (1927, 74), John Korner (1913, 8), Cornelius Krieghoff (1815, 165), William Kurelek (1927, 62), Ludger Larose (1868, 24), Fernand Leduc (1916, 13), Ozias Leduc (1864, 13), Jean-Paul Lemieux (1904, 140), Rita Letendre (1929, 80), Ernst Lindner (1897, 8), Arthur Lismer (1885, 327), Kenneth Locchead (1926, 30), Mabel Lockerby (1887, 5), Laura Muntz Lyall (1860, 28), John Lyman (1886, 19), J.E.H. Macdonald (1873, 283), J.W.G. Macdonald (1897, 48), Pegi Nichol Macleod (1904, 14), Ron Martin (1943, 9), Thomas Mower Martin (1838, 44), Henri Masson (1907, 261), Marmaduke Matthews (1837, 13), Mabel May (1884, 42), Clark McDougall (1921, 8), Jean McEwen (1923, 106), Isabel McLaughlin (1903, 17), Helen McNicoll (1879, 14), Ray Mead (1921, 18), John Meredith (1933, 22), David Milne (1882, 92), Janet Mitchell (1915, 42), Guido Molinari (1933, 21), James Wilson Morrice (1865, 15), Edmund Morris (1871, 8), Kathleen Morris (1893, 18), Norval Morriseau (1932, 149), Jean-Paul Mousseau (1927, 10), Louis Muhlstock (1904, 13), Kazuo Nakamura (1926, 39), Ernst Neumann (1907, 16), Lilias Torrance Newton (1896, 6), Marion Nicholl (1909, 11), Lucius R. O’Brien (1832, 12), Daphne Odjig (1928, 24), Will Ogilvie (1901, 6), Toni Onley (1928, 63), Paul Peel (1860, 66), Alfred Pellan (1906, 34), Sophie Pemberton (1869, 9), William Perehudoff (1919, 32), Henri Perre (1828, 5), Robert W. Pilot (1898, 293), Christopher Pratt (1935, 7), Mary Pratt (1935, 19), William Raphael (1833, 45), Gordon Rayner (1935, 8), George Reid (1860, 74), Jean-Paul Riopelle (1923, 345), Goodridge Roberts (1904, 158), Sarah Robertson (1891, 7), Albert Robinson (1881, 110), Otto Rogers (1935, 17), William Ronald (1926, 81), John Russell (1879, 15), Henry Sandham (1841, 14), Allen Sapp (1929, 118), Anne Savage (1898, 19), William Sawyer (1820, 9), Carl Schaefer (1903, 33), Charles H. Scott (1886, 11), Marian Scott (1906, 17), Jack Shadbolt (1909, 136), Gordon Smith (1919, 64), Jori Smith (1907, 55), Michael Snow (1919, 5), David Sorensen (1937, 10), William Leroy Stevenson (1905, 12), Philip Surrey (1910, 35), Marc-Aurèle de Foy Suzor-Coté (1869, 135), Takao Tanabe (1926, 52), David Thauberger (1948, 12), Tom Thomson (1877, 41), Joanne Tod (1953, 7), Fernand Toupin (1930, 53), Claude Tousignant (1932, 10), Harold Town (1924, 96), Tony Urquhart (1934, 14), Fred Varley (1881, 47), Robert Varvarande (1922, 5), Frederick Arthur Verner (1839, 90), Horatio Walker (1858, 37), Charles Caleb Ward (1831, 9), Lowrie Warrener (1900, 10), Homer Watson (1855, 30), Gordon Webber (1909, 6), W.P. Weston (1879, 110), Robert Whale (1805, 6), Curtis Williamson (1867, 7), Percey Woodcock (1855, 8), Walter Yarwood (1917, 8).