THE ART MARKET IN THE ANCIENT WORLD

Forms of art were developed since the creation of the first sedentary communities, which made it possible for some of their members to specialize in artistic decorations. Further institutional developments in the early civilizations of Mesopotamia and Egypt generated the demand for artistic commissions for temples and palaces. However, the monocratic authorities of those societies tended to promote massive constructions and the perpetuation of stylized celebratory art executed mainly by slaves and anonymous artisans rather than fostering innovations by free artists.Footnote ⁷ Some form of art trade must have taken place in these and more recent civilizations, such as the Minoic and Mycenean ones, with barter involving especially pottery, armory, and jewelry. An echo of this remains in the Iliad (VIII century BC), where Homer describes the exchange of a golden armor of the Trojan Glaucus worth a 100 oxen with a bronze armor of the Achaean Diomedes worth 9 oxen (Book VI), and elsewhere mentions the value of various awards assigned by Achilles to the winners of contests (Book XXIII).

By the end of the VII century BC, the Aegean world was composed of a multitude of independent póleis led by aristocratic institutions and connected by a florid trade network and a monetary system based on the minting of gold and silver coins. Some city-states, led by Athens, developed more inclusive political institutions where private enterprise could flourish. Inter-regional trade was expanding, and soon it was even supported by the emergence of a rudimentary credit system focused on maritime loans (Cohen Reference Cohen1992). As argued in a definitive way by Temin (Reference Temin2013), it was the beginning of a real market economy whose structure would be inherited by the larger and integrated markets of the future Hellenistic and Roman worlds. A variety of crafts developed their own markets, and we have the first evidence about art trade and prices. Before discussing this, however, we need to describe the economic and monetary system of the ancient world.

THE ORGANIZATION OF ART PRODUCTION IN CLASSICAL GREECE AND THE ROMAN EMPIRE

Art production in ancient Greece was organized by master artists through workshops. There was a system of apprenticeship for most craftsmen, including the artists, and we can reconstruct the network of masters and scholars linking most of the sculptors and painters mentioned in ancient sources.Footnote ¹⁸ Pliny explicitly mentions the case of the Macedonian painter Pamphilos of Amphipolis (active between 390 BC and 340 BC), who was paid 500 drachms a year by his pupils, including Pausias, Apelles, and Melanthios, to provide basic lessons in painting, arithmetic, and geometry (Pliny, 35, 76). As stressed by Stewart (Reference Stewart1990, p. 56), the sculptors “were entrepreneurs, operating like other crafts from a shop (ergasterion) that could be as large or as small, as fixed or as mobile, as occasion and demand required. Techne, as taught by the master-pupil system provided a solid grounding and points of future reference, but never precluded individual enterprise. In fact, the competitiveness of Greek society tended to encourage it.” And a similar argument would apply to painters, who engaged in an itinerant life across póleis to provide their services for the major public commissions.Footnote ¹⁹ In the case of complex projects, there was a real team of master artists in collaboration, as was the case for Skopas of Paros sharing the decorations of the Mausoleum of Halikarnassos with Leochares, Bryaxis, and Timotheos. All the known masters of classical Greece were freeborn (Pliny, 35, 77), and, actually, painters such as Zeuxis, Parrhasius, and Apelles were believed to own slaves (Seaman and Schutz 2017, p. 17). As in other artisanal activities, slaves could be employed, but their role must have been confined to minor tasks, as was possibly the case in Phidias’ workshops. The allocation of resources was already standardized; as noticed by Bresson (Reference Bresson2016, p. 189), “the construction of sanctuaries in the Classical and Hellenistic periods (at Epidauros, Delphi, Eleusis, the Parthenon, Delos) show that the mode of remuneration was a function of the type of activity involved. Those who provided more specialized work were paid by the piece, whereas unskilled workers were paid by the day.” In the works for the Erechtheion carpenters and sawyers were paid by day, masons and (encaustic) painters were paid by measure, sculptors, (wood) painters, and woodcarvers by piece and architects by salary, and we can guess that the contractual option was selected to optimize the incentives of the workers.

In the Roman Empire, guilds were active in various fields (Temin Reference Temin2013, pp. 108–10), but there is no evidence that they controlled prices or that they were present in artistic professions. The organization of art production was based on workshops that were mostly composed of freedmen and slaves (Stewart Reference Stewart2008; Harris Reference Harris2015). There is evidence of a workshop of mosaicists, including four freedmen, in the Albani Hills near Rome and of a workshop of painters in Castricio Street in Pompeii (Harris Reference Harris2015). Flohr (Reference Flohr2019) has estimated the demand for paintings by analyzing the distribution of almost a thousand pictures found in about two hundred buildings in Pompeii: in three-quarters of the cases, the decoration was limited to one or two rooms of the house, while only two houses have more than ten rooms with paintings. Taking into account lost works from destroyed upper floors and unexcavated parts of Pompeii, Flohr (Reference Flohr2019) estimates that the demand for panel paintings in Pompeii could have been around a hundred per year, which would justify the activity of a local workshop or at most two competing workshops in the neighborhood (and the epigraphic evidence attests to the activity of painters in nearby Puteoli and Surrentum). This evidence suggests that a city of a million inhabitants such as Rome must have hosted many more workshops, employing many freedmen and slaves in standardized decorations, which were demanded not only by patricians, but also by the middle class of merchants and shopkeepers. And the same would have happened in many other towns in all the provinces of the Empire, where remnants of similar wall paintings have been found. We also have evidence of workshops of sculptors and mosaicists active across the Roman Empire, often shipping statues and drawings for mosaics through the Mediterranean Sea, as was the case for sculptors based in Aphrodisias in Asia Minor and mosaicists whose works have been found in public and residential buildings spread around all of the Empire.

We have some information on the ordinary activity of the artists, their costs, and their contractual agreements. Bronze statues were commissioned at prices that internalized the cost of metal, but left space for adjustments: classical Greek sculptors used alloys with mere traces of lead and an abundance of tin, which was more expensive, while later Hellenistic and Roman artists often decreased the percentage of tin, used thinner walls to reduce the amount of metal used, and introduced other cost-saving techniques.Footnote ²⁰ Paintings were originally based on a few colors, and classical Greek painters gradually introduced new varieties of colors; instead, Roman painters introduced cheaper versions of the most expensive colors, such as a lower-quality blue (Ling Reference Ling1991, p. 209). Vitruvius and Pliny inform us that the patrons were usually paying for expensive colors, which were the so-called “florid” pigments, while the painters were responsible for the cheaper “dry” pigments. Nevertheless, more flexible arrangements were also possible: Ling (Reference Ling1991, p. 217) mentions a Greco-Egyptian papyrus of 255 BC where a painter proposed a price of 30 drachmae excluding costs or, in alternative, a price of 53 drachmae including costs for the same work: We do not know what was the final choice of the patron, but basic principles of contract theory suggest that incentives to deliver high-quality work would have been stronger in the former case and incentives to reduce costs in the latter.

Painting techniques of the Greco-Roman world included secco and fresco painting, tempera, and encaustic (hot wax) painting. Easel paintings were typically executed by a single master in the workshop. Instead, wall paintings were usually executed on the spot by a team, which in the Roman world included an organizer of the works defined as pictor et praepositus pictorum, a leading painter for the main figurative works named pictor imaginarius, a painter specialized in landscapes and decorations named pictor parietarius, a plasterer and assistants. While Greek art developed the individual production of easel paintings also for the most exclusive environments, which fostered by its nature the development of a secondary art market, Roman painting was almost exclusively wall-painting carried out on plaster and covering the entire walls of private houses, which limited the development of a secondary market. Moreover, the best painters were commissioned specific works in Greece and paid for them, while payments per day were common for the painters of the Roman world: the difference again is crucial, since direct competition for sequential commissions tends to incentivize quality and artistic innovation, while the provision of labor paid on a daily basis tends to penalize quality and incentivize repetitive tasks, such as the replica of the same or similar subjects on multiple frescoes (also using the same drawings).

Classical Greek art was largely focused on idealized mythological and historical subjects. Hellenistic art started to diversify the supply toward other subjects, but it was Roman art that exploited all the opportunities for product differentiation, in particular in painting. Large figurative paintings, called “megalographic paintings,” were commissioned by temples and other public buildings in ancient Greece, and the tradition continued in Rome, also for domestic decorations and largely through replicas of classical compositions. Besides this, Roman art developed further the specialization in the minor genres, from portraits (realistic and also for the lower class, as was the case in sculpture), to landscapes (including cityscapes, marine and harbor views, and more), genre paintings (including scenes of daily life, as well as burlesque and erotic subjects), and still life paintings (including fruits, flowers, birds, fishes, dead animals, and other trompe-l’œil subjects). As in other subsequent periods of art history (Montias Reference Montias2002; Etro, Marchesi, and Pagani Reference Etro, Marchesi and Pagani2015), specialization in genres and also in particular subjects became an efficient way of allocating work within workshops, generating scale economies in repetitive tasks.

The economic and social background of the master artists was different between classical Greece and the Roman Empire. In Greece, it was associated with the middle class of free citizens, as shown by the mobility of the master artists across city-states and their occasional elections to priesthoods and magistracies. We also have evidence that famous artists were close with intellectuals, and they often interacted with the elite. Polygnotus was the lover of the sister of the Athenian statesman Kimon; Phidias was a good friend of Pericles and the philosopher Anaxagoras; and Praxiteles’s family married into the Athenian aristocracy. Many artists wrote treatises on art theory. Parrhasius mingled in Socratic circles, and the same Socrates belonged to a family of sculptors and had probably been trained in the art. Ironically, speaking in the dialogues of Plato (as the Republic, written around 380 BC), Socrates excludes the arts from the ideal State because they just involve imitation of nature. However, the wide space dedicated by Plato to art in his writings suggests that he was not considering it as any other craftsmanship. Aristotle moved to the court of Philip II of Macedon in 343 BC to tutor the son Alexander, and there he must have been in close contact with leading artists such as Lysippos and Apelles, who were treated as peers by the same Alexander. Indeed, Aristotle had a much more positive view of the arts in his Poetics, written around 335 BC. The social status of the artists changed in the Roman world.Footnote ²¹ The negative connotation associated with sculpture was a concrete burden: Plutarch claimed that no noble young, upon seeing Zeus at Olympia or Hera at Argos, would desire to be either a Phidias or a Polykleitos, and Lucian was expressing a similar concept in the middle of the II century AD. Most of the tombs of artists in Italy refer to freedmen or eastern slaves, and important families had slaves who were specifically employed in services concerning handicrafts and painting (Digest, XII, 6, 26, 12). Patricians and intellectuals would occasionally paint as amateurs, but the professional workshops were populated by anonymous craftsmen. The contrast between classical Greece, where market forces led artists to work where they would get the best deals, and the Roman world, where artists produced works to be traded in the open market, was particularly clear in sculpture. A lively piece of evidence comes from a novel written by Philostratus around 238 AD about the Life of Apollonius of Tyana, a philosopher of the I century AD:

Theoretical Predictions

We have emphasized major differences between art production and innovation in classical Greece and during the Roman Empire. While these periods had much in common in terms of cultural values and socio-economic organization, it is also clear that multiple factors may have shaped the different evolutions of artistic creativity, and an empirical analysis cannot identify the role of all these factors, let alone the causal components, with the limited data available from ancient markets. Our more modest goal is therefore to emphasize a structural difference between art markets in classical Greece and the Roman Empire, advance a plausible mechanism through which it may have affected the paths of artistic evolution, and verify whether its implications for art pricing are consistent with the evidence. The structural difference is about the size of the relevant art markets, and the mechanism is about the kind of artistic production and innovations that markets of different sizes could promote. Theories of cultural economics, starting with Cowen and Tabarrok (Reference Cowen and Tabarrok2000), have often emphasized that market size affects the incentives to generate art that appeals to elite vs. mass tastes and contributes to determining the kind of innovations in which artists invest. Small markets with demand driven by a few wealthy patrons for original works tend to promote investment in quality-enhancing innovations to reach high price commissions, while large markets with demand driven by a mass of anonymous buyers tend to be biased towards reproducibility, which encourages cost-saving innovations and discourages originality. Here we provide a similar insight with a stylized model of artistic evolution.

Let us consider a representative workshop investing in a quality-enhancing activity and a cost-reducing activity, denoted respectively with the Greek letter ϵ _t and the Latin letter i _t at time t. In each period, the workshop bears a positive sunk cost F(i _t), which is increasing and convex in the investment aimed at reducing the marginal cost, for instance, to set up a larger factory, employ more slaves, or find new ways to reproduce art at a low cost. The unitary cost of an artwork C(ϵ _t, i _t) is increasing and convex in the quality-enhancing investment, for instance, aimed at technical, stylistic, and compositional innovations, and decreasing and convex in the cost-saving investment. The demand is reflected in a price function, P(q _t) which represents a willingness to pay, increasing and concave in the quality q _t of artworks as perceived by patrons. The equation of motion for artistic quality depends on the accumulation of innovations transmitted over time through apprenticeship within workshops, or:

where λ ϵ [0,1] is a rate of quality transmission net of obsolescence of old techniques and traditions. The market size L determines how many artworks are sold, generating the profits of the representative workshop:

to be maximized with respect to ϵ _t and i _t in each period, taking as given past quality. We denote derivatives with subscripts and assume C _ϵi (ϵ _t, i _t) ≥ 0, so that cost-reductions tend to be harder for higher quality art. Assuming interior solutions, the equilibrium investments satisfy:

The first condition equates the marginal benefit of the investment in quality to the marginal cost per artwork, and the second equates the marginal benefit of cost reductions over all sales to their marginal cost. Together, they determine the evolution of art quality q _t , which, under standard regularity conditions, converges to a steady state with stable quality q˜ and delivers the comparative statics ∂ϵ _t /∂q _t–1 < 0 and ∂i _t /∂q _t–1 ≥ 0, implying the stability condition ∂q _t /∂q _t–1 < 1. In such a context, the evolution of art history is initially associated with major artistic innovations aimed at monetization by rapidly increasing prices, which then implies decreasing quality-enhancing investments and increasing cost-saving investments while art prices converge to a stable level. More important for our purposes is the comparative statics with respect to the market size L, implying ∂ϵ _t /∂L ≤ 0 and ∂i _t /∂L > 0 under our assumptions. A small market size generates low investments in cost reductions and high investments in quality enhancement, with prices increasing at a fast rate, especially when the willingness to pay is particularly sensitive to quality. Instead, a large market size stimulates investments in cost reductions because these are spread through more sales and decreases investments in quality enhancement, speeding up convergence to stable prices across an integrated market.

Empirical Implications

The classical Greek world was small but rich. Estimates discussed in Bresson (Reference Bresson2016) suggest a population of 200/250 thousand inhabitants in Attica, 3 million people on mainland Greece (including Macedon), and less than that in Greek-speaking towns across the Aegean Sea and other colonies. The artists could serve the demand of the main city-states, with patrons who were directly competing to attract the best works within a relatively small region. Given the public nature of Greek paintings, and the fact that they were hard to reproduce cheaply (and had a resale value), this market was biased in favor of elite tastes. And taste for more elaborate, higher-quality paintings was growing over time (because of competition for prestige, growing patron demand, and learning by the artists); this led to an increase in the derived demand for specific skills by top artists and in their earnings. The subsequent Hellenistic world represented a sort of trait d’union between the Greek and Roman worlds and was characterized by an expansion of the size of the market and a gradual emergence of workshops focused on vast production for wide commercialization. The Roman Empire developed an even wider and well integrated market. Estimates discussed in Temin (Reference Temin2013) refer to a population of 55/70 million people. In such a context, large workshops could serve the demand of an anonymous mass market distributed around the entire Mediterranean Sea by either shipping their products or migrating where there was more demand.

Different empirical predictions emerge for our two main periods. As we have emphasized, the rich communities of the Greek póleis competed to obtain artworks of higher quality, which fostered the entry of new artists and, most of all, competition for the development of a series of product innovations (new techniques and novel ways to create illusionistic painted and sculpted images). While public and (major) private commissioners were outbidding each other to attract the best artists, new heights of artistic creativity were reached, and the market value of masterpieces had to increase over time. Our empirical analysis in the next section verifies this hypothesis by focusing on major paintings by top artists for leading commissions in classical Greece, on which we have records between the V century BC and the beginning of the III century BC. The fact that these are famous works by famous artists (literally) surviving the test of time through their mention in literary sources (even if the actual works did not survive) is what makes our dataset consistent with our focus on the maximum achievements of artistic creativity. Similar conditions appear in later periods of art history when competition among local elites fostered increasing art prices and a steady flow of quality-enhancing innovations, as notably during Renaissance Italy with competition between patrons of different towns, in London at the end of the XVIII century with competition between rich collectors, and in Paris after the liberalization of art exhibitions at the end of the XIX century with competition between rival dealers.

The Roman Empire featured a much larger and integrated economy where all the provinces around the Mediterranean Sea were connected by safe maritime trade. The demand for artworks derived essentially from all the provinces, in spite of their heterogeneity in wealth and traditions, and this promoted competition to serve this demand by shipping works or directly migrating where the works were needed, fostering process innovations (cost-reducing techniques, employment of slaves, copying, and product differentiation). Cost-saving investments for mass production were more profitable in a large market than investments in quality-enhancing innovations. This should have led the market value of traded artworks to be stable (or decreasing over time as a consequence of efficiencies in production). Most of all, the prices for similar artworks should have been largely equalized across the different provinces; otherwise, market forces would have eliminated the remaining price differentials. Our empirical analysis will verify these hypotheses on statues commissioned between the I and III centuries AD in towns distributed across the Roman provinces. The information derives from inscriptions found at the bases of the statues, while the actual statues were subsequently removed or lost. The patrons could be either public institutions or private donors from various provinces. The fact that these are inscriptions that survived the test of time (even if the statues and the names of the artists did not) across a huge territory appears consistent with the idea that the sample is representative of these commissions. Also in this case, we should mention that similar conditions appear in later periods of art history when large art markets led to mass production, featuring stable (or decreasing) art prices and a continuous process of entry of artists engaged in product differentiation and cost-saving innovations, as in the Dutch golden age with a new demand of paintings by the middle class, or in Paris during the early XX century with a new international demand for modern art.

EVIDENCE ON THE PRICE OF PAINTINGS IN CLASSICAL GREECE

The earliest records for commissions of paintings in ancient Greece concern the primary market. We already mentioned works depicting episodes from the Iliad and an Amazon battle in the Stoà Poikile of Athens around 470/460 BC. While the first one by Polygnotus was not paid, we know that Mikon was fined half a talent for depicting the Persians as larger than the Greeks. Assuming that the composition of the battle was equally divided between the two armies and that the punishment was proportional to the original payment, it does not seem unreasonable to conjecture that a talent could have been the price initially agreed for the entire composition. This would also resonate well with the order of magnitude of the compensation received by Polygnotus for his earlier commission after the battle of Plataea.

Pliny mentioned two paintings sent by Zeuxis as gifts to the town of Akragas and to the king of Macedon, which we can temptatively date around 410 BC. Also in this case, the artist was probably looking for a subsequent return, which did arrive. According to Aelian (Varia Historia, 14.17), the painter went to the Macedonian court of Archelaus I for a decoration of his palace at Pella and was paid 400 mines for a work, the first clear record for the payment of a painting in history. This could be temptatively dated around 400 BC, given the information by Pliny on when the painter reached his acme. The price was about 6.6 talents, and since Zeuxis was mainly a panel painter of few figures, we can conjecture that the payment must have been for an easel painting. Later mosaics left in Pella, typically representing at least a couple of human figures, may have reflected the style or even the composition of such a painting (Moreno Reference Moreno1987).

One of the best-documented Greek painters in terms of transactions is Aristeides of Thebes, who was active around 360–310 BC. He painted a “Battle with Persians” featuring a hundred men, each one paid ten mines by Mnason, the tyrant of Elatea between 338 and 297 BC, for a total of 1,000 mines or 16.6 talents (Pliny, 35, 99). The same Mnason purchased two other paintings, agreeing on a fixed price per figure. A painting of mythological “Heroes” by Theomnestos was paid 20 mines per hero. A painting of “Twelve gods” by Asklepiodoros of Athens, who was rated as the major Athenian painter of his age by Plutarch and was admired by Apelles, was paid 30 mines per deity by Mnason, for a total price of 6 talents. The number of heroes was probably intermediate among the 12 gods and 100 men, but closer to the former given the available number of heroes in Greek mythology: a total price of 10 talents would emerge, conjecturing 30 heroes. According to Pliny, Aristeides and Asklepiodoros flourished around 332–329 BC, which could be the period of these three commissions from Elatea. It is also possible that the paintings, at least the two by Asklepiodoros and Theomnestos, which were jointly mentioned by Pliny, were also of similar size to be complementary in the decoration of the same room in the palace of the tyrant. The payment per figure was a natural way of incentivizing effort based on a quantifiable feature of the painting, as it will be in later centuries (see Etro and Pagani Reference Etro and Pagani2012). Moreover, the price per figure was directly related to the intrinsic status of the figures (gods, heroes or semi-gods, and humans), but also inversely related to the total number of figures, which suggests a quantity discount.

For the most celebrated Greek painter, Apelles, active around 350–310 BC, we know many anecdotes, occasionally including price matters. Alexander the Great paid 20 talents for his portrait as “Alexander Keraunophoros” or wielding a thunderbolt, a distinctive attribute of Zeus (Pliny, 35, 92). The work was destined for the Temple of Artemis in the newly conquered town of Ephesus. Since the reign of Alexander the Great lasted from 336 BC to 323 BC and the Temple of Artemis was probably completed at a late moment, we can date the work after the commissions of Mnason and during the Persian campaign of Alexander, probably around 325 BC, when the Porus dekadrachms featuring Alexander with a thunderbolt were struck. Another anecdote by Pliny reinforces the message that Apelles could obtain higher prices later on, temptatively in the last two decades of the IV century BC. When Apelles visited Protogenes in Rhodes and asked about the price of his works, Protogenes mentioned a small amount. Then, Apelles offered 50 talents for a painting and spread the word that he was going to resell it as his own painting. This signaled to the Rhodians the high esteem he had for the rival, and Apelles ended up selling the paintings at prices higher than what he paid (Pliny, 35, 88). We can guess that the prices of paintings by Protogenes, a sort of experimental artist in the sense of Galenson (Reference Galenson2024), must have been increasing in the last phase of his career. Most important for our purposes, what we can reasonably infer from this anecdote is that at that time, arguably around 315 BC, Apelles was able to sell a painting for 50 talents, though we cannot draw any conclusion on the size and composition corresponding to such a high price. This shows not only the business acumen of a leading artist such as Apelles, engaged in an early case of successful art investment, but also how quickly price increases could spread around Greece.Footnote ²³ At the end of his life, Apelles moved to Kos, where, between 306 and 301 BC, he painted one of his most famous paintings, an “Aphrodite Anadyomene” (Venus rising from the sea), left unfinished for the temple of Asklepios. We will find this painting again, traded in the secondary art market.

According to Plutarch (Moralia, 1093), King Ptolemy I of Egypt offered 60 talents to Nikias of Athens for a “Nekyomanteia,” or consultation of the souls, drawn from the Odyssey of Homer. Nikias was one of the most recognized history painters of the time, and by then he must have been old, rich, and famous since he refused the sale and donated the painting to his own town. In Book XI of the epic poem, Odysseus spoke with Elpenor, Tiresias, and Anticleia on his underworld journey and met dozens of souls, which suggests that this was a wide composition with a large number of figures, comparable to (and possibly inspired by) the one of Polygnotus on the same subject. Since the reign of Ptolemy I lasted from 305/4 BC to 282 BC, we can temptatively date the painting to 290 BC. In any case, this is the last and also the highest documented price evaluation that we find in the primary market for ancient Greek paintings.

Further information is from the secondary art market. In the II century BC, King Attalus of Pergamon bought a painting by Aristeides of Thebes for 100 talents (Pliny, 35, 100). This was probably Attalos II, whose reign lasted from 159 to 138 BC, exactly the time in which Pliny argued that art was revived after a period of stagnation. He was a renowned patron of the arts (and completed the Great Altar of Pergamon) as well as an avid collector of works by Aristeides, since he was again interested in a painting by him at an auction in 146 BC, which gives us the first record of an auction sale of a precise painting. According to Pliny, the auction took place after the sack of Corinth by the Roman army of Lucius Mummius, and the painting was a “Dionysus and Ariadne” by Aristeides confiscated by the Corinthians after the battle. However, the sale to Attalos for the same amount of 100 talents surprised Mummius so much that he wanted the painting back and brought it to the temple of Ceres on the Aventine in Rome (Pliny, 35, 24). Since this happened at the end of the reign of Attalos, we can conjecture that he already owned the (probably) distinct painting by Aristeides purchased at the same price, and that size and composition were not too different, justifying a similar expenditure. Given the wide and long career of this painter, possibly another experimental artist, it seems plausible that these masterpieces traded in the secondary market were from the late activity of Aristeides, which took place between 320 and 310 BC.

We are also aware of transactions of lower value in the later secondary art market of Rome. In the first half of the I century BC, the Roman orator Hortensius purchased a panel of “The Argonauts” by Kydias for 6 talents, and we are told that he built a small shrine for it in his villa of Tusculum (Pliny, 35, 130). The destination and the title suggest that the painting was large and contained more than a couple of figures, but the subject was probably the pretext for a landscape or marine painting, and therefore the figures may not have played a dominant role in the composition. For a cheaper acquisition, we can look at copies of masterpieces. Pliny mentions a I century BC copy by Dyonysios of Athens of a famous painting by Pausias, which was therefore originally executed around 350 BC. The copy was purchased for 2 talents by the Roman statesman Lucius Lucullus (118–56 BC) at the Dionysia festival in Athens, possibly in 88/87 BC, when he was documented there. It was a portrait of a young girl named Glycera, a lover of Pausias, selling garlands of flowers (Pliny, 35, 125). This is the oldest painting with a content of portrait and still life for which we have records of a commercial transaction, and its subject would have been perfect in a “lucullan” banqueting room. Taking into account the lower position of still life in the hierarchy of paintings and the fact that this one was just a copy, the price was remarkable. It is probably not a coincidence that Pliny, with his encyclopedic attitude, managed to provide price records for masterpieces in all the main genres. And it is not unreasonable to assume that his small sample of price records was representative of the larger sample of masterpieces he must have read about (in Greek sources) while preparing the chapters of the Naturalis Historia on art.

Julius Caesar was also a collector. Though his military campaigns in Gaul and his forays into Britain and Germany could not find artworks comparable to those of the Eastern world, according to Suetonius (Life of Caesar, 47), “he attacked Britain in the hope of finding pearls… and he was always avid collector of gems, engravings, statues, and paintings of ancient workmanship.” After his victory over Pompeus at Pharsalus in 48 BC, Caesar personally purchased two paintings by Timomachus of Byzantium, an “Ajax” and a “Medea,” for 80 talents to be placed in the Temple of Venus Genetrix (Pliny, 35, 136). The same painter may have been the author of an “Ajax” and an “Aphrodite” purchased in Cyzicus, near Byzantium, for 50 talents by Marcus Agrippa (63–12 BC), a statesman promoter of the necessity of exhibiting art in public collections rather than in private villas (Pliny, 35, 26). Pliny reports that Timomachus left “Ajax” and “Medea” unfinished when he died, which may have been right before the purchase by Caesar. However, there are good reasons to doubt that Timomachus was a painter of the I century (see, for instance, Gutzwiller Reference Gutzwiller2004). It seems unlikely that, in a period in which many artists were migrating to Rome (including Iaia of Cyzicus), a major artist was still active in the periphery of the Roman world. Most of all, already in 70 BC, Cicero had pointed out the existence in Cyzicus of two masterpieces representing “Ajax” and “Medea,” wondering how much they could cost (In Verrem, IV, 60, 135). It seems likely that these masterpieces were those purchased by Caesar for such a prestigious destination in Rome, which would suggest that the artist was not a contemporary painter, but a famous master of earlier times. This has supported the hypothesis that Timomachus was active in the early III century BC or even in the late IV century BC. We will endorse this convincing hypothesis and temptatively assign these works to 300 BC.

Octavian Augustus was not an art collector and did not decorate his own house with statues or paintings, but was nevertheless interested in bringing classic masterpieces to Rome for public display. The contemporary historian Strabo claims (Gheographikà, 14, 657) that Augustus purchased the “Aphrodite Anadyomene” of Apelles from the temple of Asklepios in the city of Kos, remitting 100 talents of due taxes. The same interest may have been shared by the cultural administrator of Augustus, the wealthy Gaius Clinius Maecenas, who became famous as a patron of the arts and a supporter of poets such as Virgil and Horace. However, there is no evidence that such “maecenatism” extended to contemporary painters and sculptors, another sign that at the time the local artists were mainly artisans without a relevant role in the cultural life of Rome. The emperor Tiberius was another avid collector who owned works at least by Zeuxis, Parrhasius, Apelles, Protogenes, and Nikias and had a special interest in lascivious subjects. He destined to his private rooms an “Atlanta and Meleager” by Parrhasius that evaluated 40 talents,Footnote ²⁴ and the famous statue of the “Apoxymenos” by Lysippos.

Under the heroic assumption that we can turn into data the conjectures made earlier, we can put together 21 price records. In Figure 1, we display these prices corrected for the purchasing power of a talent in 450 BC in Athens based on the evidence of wage inflation in Loomis (Reference Loomis1998). The estimated date of execution covers almost two centuries, and we distinguish prices from primary sales in solid circles and from secondary sales in empty circles. The amount of data is of course limited, but an increasing trend in real prices emerges overall and also, focusing only on sales in the primary market, with the top prices for works executed at the end of the IV century. A more formal empirical analysis confirms such a positive trend in hedonic regressions, controlling for additional information available on these artworks. In particular, we have used different controls for the quality of the artists. In one case, we used dummies for artists with multiple observations, that are Zeuxis, Apelles, and Aristeides. While the first is associated with a negative coefficient and the last two with positive coefficients, the inclusion of their dummies does not affect the evidence of a positive time trend. In other specifications, we have employed a proxy of artistic quality as perceived in ancient times based on the length of the biography of each painter in the Naturalis Historia of Pliny. Such a Pliny index contributes to explaining price variability, providing a positive and significant coefficient, but does not eliminate the significance of the positive time trend.Footnote ²⁵ The estimated number of figures in the paintings has always a positive impact on prices, but this is neither economically nor statistically significant. The dummy for sales in a secondary market has a positive and sometimes significant coefficient. The unconditional time trend corresponds to an increase of 1.6 percent per year in the real price of masterpieces, and remains above 1 percent per year in full specifications and in robustness checks limited to sales in the primary market or sales at positive prices. Overall, the evidence supports an increase in the price of masterpieces between the beginning of the V and the end of the IV centuries BC, which is when the literary sources document a sequence of major artistic innovations.

FIGURE 1 REAL PRICES OF PAINTINGS IN CLASSICAL GREECE, V–III CENTURY BC

Notes: The prices of the primary market are in solid circles and the prices from the secondary market are in empty circles. The year refers to the estimated date of execution.

Sources: Nominal prices from literary sources. Adjustment for purchasing power from Loomis (Reference Loomis1998).

EVIDENCE ON THE PRICE OF STATUES IN THE ROMAN EMPIRE

The main piece of evidence on art prices during the Roman Empire comes from inscriptions at the base of statues spread around all the provinces, sometimes recording prices of the commissions to honor the munificence of the donor. Since towns that have been repeatedly built during the following centuries are less likely to conserve such remnants, a large part of the epigraphic evidence derives from peripheral towns and archeological sites that have not been excavated until recent times, whose distribution covers most of the “Romanized” provinces of Europe and Africa, where many colonies founded or populated by Roman citizens were being decorated with temples, forums, baths, and villas that required new statues. The wide geographical distribution of the evidence is quite important because our interest is in comparing prices across different and remote locations of the Empire rather than identifying prices of works ordered in Rome by emperors under unique monopsonistic conditions. The statues were commissioned by local communities and private donors for public display, and they were erected on bases engraved with information on the commission, its subject, and its purpose. They included statues of gods, heroes, and portraits of the emperor (often replicas of models sent from Rome), portraits of women of the imperial family, and portraits of the same donors.

The first collection of records of prices of statues was put together by Friedlander (Reference Friedlander1913), but the most comprehensive one is due to Duncan-Jones (Reference Duncan-Jones1982), who focused most of his attention on the African provinces, which covered a large area between modern Morocco and Libya. At the time, these provinces were Mauretania, Numidia, and Africa Proconsularis (Byzacena, Tripolitania, and Zeugitana), including Numidia Procunsularis. In these provinces, the median price of statues was HS 5,000, around six times the average annual labor compensation in the early Empire (Temin, Reference Temin2013). But the variability was also high: the major town of Leptis Magna left a statue priced at HS 460 and a commission of 16 statues for a total of a million sesterces. As an example, focusing on dated standalone marble statues of Septimius Severus erected in Numidia at the beginning of his reign between 193 and 198 AD, we find prices of HS 2,400 and 3,400 in a rural center of modern Algeria and HS 4,000, 4,300, and 10,000 in the nearby town of Diana, suggesting a certain variability.

Italy provides few price records analyzed by Duncan-Jones (Reference Duncan-Jones1982, p. 126), who admits that his sample is too small and uneven to draw solid conclusions on price comparisons, but suggests that “it is reasonable to suppose that the norms in Italy would not have been very different from those in Africa, where the heaviest concentration of statue prices lies in the range from HS 4,000 to HS 7,000.” Also in Italy, price variability was high, in a range including a statue from Volsini in Etruria priced HS 500 before 200 AD and a silver chariot of the II century priced HS 100,000 in Formiae in Latium. We have added to the dataset a few more records from Italian regions whose inscriptions were published during the last four decades. For instance, this was the case of Barium in Apulia with a statue of HS 10,000 from the middle of the II century (AE 2008, 416), Misenum in Latium with HS 20,000 in 161–211 AD (AE 1995, 311), Potentia in Lucania with HS 10,000 during the I century (AE 1995, 370), and Interamnia Praetuttiorum in Picenum with HS 4,000 during the II century (AE 2013, 194). The final dataset includes statues from the regions of Latium and Campania, Lucania, Venetia, Umbria, Apulia, Etruria, Transpadania, Aemilia, Picenum, and Samnium.

Gallia and other northern European provinces provide an additional group of price records. Drinkwater (Reference Drinkwater1979, Reference Drinkwater2014) has mentioned a couple of statues from Aventicum in Germania Superior, to which we should add a statue of Mars priced at HS 48,000 at the end of the II century in Mutigney (in modern Swiss territory) for a testamentary outlay (AE 2014, 896). The nearby province of Gallia Narbonensis left a bronze of HS 6,000 in Arles (AE 2002, 921) and silver statues from Nimes (AE 1982, 682, and CIL 12, 3058), Narbonne (CIL 12, 4445), and Vienne (CIL 12, 5864). From other northern provinces, we also have a couple of price records in Pannonia Inferior (in modern Hungary), respectively HS 50,000 for a statue from Aquincum (AE 1990, 810) and HS 40,000 for a statue from nearby Vetus Salina (AE 2013, 1201). We should also add the unique case of a Roman artwork that arrived to us with the name of the artist and price, a bronze statuette from Britannia whose base has inscribed a dedication to Mars from the Colasuni brothers: they paid HS 112 to the coppersmith Celatus, of which HS 12 was for the bronze material and the rest for the execution. This corresponds to a cost of artistic manufacture representing about 90 percent of the total price, probably in line with common works in marble, but higher compared to larger metal statues (for which the cost of the material could be more relevant).

Spain was deeply “Romanized,” and its local communities ordered many statues for which we have price records. Curchin (Reference Curchin1983, Reference Curchin2014) has collected a list of prices of statues for the Spanish provinces that we use in our dataset, with a few additions. In particular, we add a statue valued at HS 7,505 according to an inscription from Guadix in Hispania Citerior (CIL 2, 3390) and a price of HS 4,800, probably for two statues commissioned in Hispania Baetica (CIL 2, 1359). We have aggregated observations from these provinces under the group from Hispania.

Asia Minor provides more interesting evidence added to our dataset. Barresi (Reference Barresi2003) has analyzed its commissions, with a particular focus on the province of Asia. He has pointed out seven statues representing the different tribes of the town of Philadelphia, which were paid a total of HS 28,000 at some point after 212 AD, and a statue representing its own donor, Aeliua Aelianos, recorded in the town of Thyatira and paid HS 4,000 between 222 and 235 AD. Moreover, the major town of Ephesus has provided epigraphic evidence of a few groups of silver statuettes dated in 104 AD. For six of them, each one representing Diana Artemis of Ephesus, together with two other figures, we have the prices, ranging between HS 6,533 (IK-11-01, 34), HS 17,000 (CIL 3, 14195-4, and HS 33,000 (CIL 3, 14195-5,6,7,8). More limited is the information from Greece. However, we can exploit a literary source from the early II century AD: Dio Chrysostom (Discourses, 31, 59) mentioned that at the time one could erect statues at the cost of HS 2,000 or HS 4,000 in Rhodos, which also belonged to the province of Asia. Finally, a small marble statue was priced at HS 140 during the II century AD in Philippi within the province of Macedonia (CIL 3, 633). We group these observations from Greece and Asia.

The final dataset includes 202 price records (Etro 2023). The basic information contains the price and the place of the commission. Between statues from Italy (23 observations), we have most of the records from Latium and Campania, Lucania, Venetia, Umbria, Apulia, and Etruria. Outside of Italy, the province featuring most of the prices of single statues is Numidia (73 observations), followed by Africa Procunsularis (66 observations), Asia, Hispania Baetica, Gallia, Mauretania Caesariensis, and Germania Superior. For about two-thirds of the observations, we can recover from the inscriptions the date of commission of the statue or, more often, the approximate period. On this basis, we have allocated statues between those from the I century (including the Julio-Claudian and Flavian dynasties), the period of the emperors Trajan and Hadrian (98–138 AD), the Antonini emperors (138–192 AD), including Antoninus Pius, Lucius Verus, Marcus Aurelius, and Commodus, the dynasty of the Severans (193–235 AD), including Septimius Severus, Caracalla, Geta, Elagabalus and Severus Alexander, and the Mid III century. When the estimated period of commission overlaps between two of these groups, the statue is attributed to both of them. For quantitative purposes, we have also assigned the exact year of execution when this was available or an estimate when the range was well defined in a limited period (123 observations).

Prices are all reported in sesterces (possibly converted from denari), and they all belong to the period preceding the beginning of inflationary pressures in the last three decades of the third century. Table 1 reports descriptive statistics for the dataset. In almost all cases, we know the subject of the statue. Portraits of Roman emperors are common in the dataset (35 observations), and Hadrian is the most frequent one, which is consistent with the fact that there are more surviving portraits of him than of any other emperor after Augustus: statues of Hadrian are recorded in Africa Procunsularis at Hippo Regius in 117/138 AD for a price of HS 17,000, at Vina in 138/161 AD for HS 2,400, and at Suturnuca in 146 AD together with a statue of Lucius Verus for a total price of HS 5,525. Other frequent emperors in the dataset are Antoninus Pius, Marcus Aurelius, Commodus, Septimius Severus, and Caracalla. For instance, Antoninus Pius appears in Samnium with a statue of HS 4,000 dated 140 AD, in Mauretania with one of HS 6,000 dated 156 AD, in Africa Procunsularis with two statues priced at HS 600 and HS 800 around the same period, and elsewhere jointly with other emperors.Footnote ²⁶ There are also female imperial portraits such as those of Julia Domna found in Ammaedara and Medeli in Africa Proconsularis, priced at HS 10,000 and HS 3,000, respectively, at the beginning of the third century. Another frequent subject is represented by mythological gods and heroes with 55 statues, often about Jupiter (present, for instance, with two statues dated before 200 AD and priced at HS 800 in Venetia and HS 10,000 in Transpadania, one of HS 6,000 from the II century in Baetica, and two statues erected in Numidia in 182 and 212 AD for HS 5,000 and HS 4,000, respectively), Heracles (with an early statue from Umbria priced HS 30,000 and others from Mantua as well as Numidia), and Aesculapius. Other frequent gods are Mercury, Apollo, Iuno, Minerva, Mars, Pluto, and Neptune. We also have portraits of donors (23 observations) and many idealized figures, such as the youthful male Genius Populi Romani or the female Victoria Augusta and Fortuna Redux.

TABLE 1 DESCRIPTIVE STATISTICS ON ROMAN STATUES, I–III CENTURY AD

*Statistics on real prices and year are for commissions with estimated period of execution (123 observations).

Sources: Our elaboration on nominal prices from epigraphic sources (202 observations). Adjustment for purchasing power from Wassink (Reference Wassink1991).

Our records are confined to cases where we know the exact number of statues commissioned for a given price: most of the records refer to one statue, but there are many instances of a couple of statues and cases of multiple ones. We often have information about the material of the statues, and a conjecture by Duncan-Jones (Reference Duncan-Jones1982) is that when the material is not specified in the inscription, the statue is likely to be in marble. While this conjecture cannot be verified, it was probably correct in a majority of cases; therefore, we consider marble and unknown material as our baseline category and build dummies for statues in bronze (9 observations) and silver (17 observations). We occasionally know when the price also covered an aedicola (we excluded more complex constructions as those combining arches and small temples with a statue) or the same base, and when it was referred to an equestrian statue (as in two portraits of Septimius Severus erected in 196/197 AD for HS 12,000 or the one of Marcus Aurelius erected in 146 AD for HS 8,000, all of them in Zeugitania), and we classify these as cases of large statues (38 observations). Instead, statuettes and busts are classified as small statues (23 observations). In some cases, we know the purpose of the commission, as in the 32 cases of testament outlays (typically associated with private commissions after the death of the donor or a relative), and occasionally we also know the name of the patron who paid for the commission. Since we know the exact provenance of the statues, we have identified those commissioned in large towns (48 observations) and coastal towns with direct access to the Mediterranean Sea (26 observations): in the former case, one may expect higher prices due to higher demand or richer patrons, and in the latter, one may expect lower prices due to lower costs of shipping statues (or better access to information due to connection with the rest of the world) compared to inland and remote towns.

Figure 2 reports prices in sesterces corrected for the cost of living in the Augustean age (early I century) using the index of Wassink (Reference Wassink1991) and is limited to commissions with precise dates of execution. Overall, the commissions cover more than two centuries and emphasize a great variability of prices around a median price of HS 5,000 in nominal terms and HS 3,462 with adjustment to the price level in the Augustean age. The statistics on the nominal prices by province and subject confirm substantial variability, with prices in northern Europe, Africa Procunsularis, and Italy that are above those of Spain and other African provinces. However, a large part of this variability can be explained by the characteristics of the underlying commissions.

FIGURE 2 PRICES OF STATUES IN THE ROMAN EMPIRE, I–III CENTURY AD

Notes: The prices are in sesterces adjusted for changes in the cost of living relative to the Augustean age. The years refer to the date of dedication of the commissioned statues. The date is estimated when the period of execution is uncertain (over less than a century).

Sources: Nominal prices from epigraphic sources. Adjustment for cost of living from Wassink (Reference Wassink1991).

The regressions of Tables 2 and 3 on nominal and real log prices are divided between a version with basic characteristics of the statues in Column (1), a version controlling for temporal differences in Column (2), one augmented also with destination dummies in Column (3), another also including the subjects of the statues in Column (4), and finally the full specifications. Table 2 shows that the nominal prices covary in an expected way with the number of statues (the relation is increasing and possibly concave), and with their size, as well as with the use of silver, which is priced above the omitted material (mostly marble), and above bronze. The price appears higher than average in case of testamentary outlays. The subject of the statues does not appear to significantly affect their prices, as one should expect in a competitive market where artists are employed on different subjects. Moreover, nominal prices do not change much over time, though there is a slight increase in the latest periods, which, as we will see, is in part due to the underlying inflation. Consistent with what one may expect, coastal cities are associated with lower prices and large cities in each province with higher prices; however, the coefficients of both these control variables are not statistically different from zero. Without controls, statues for Italian locations exhibit higher than average prices, and those for Greece and Asia Minor exhibit lower than average prices. However, in the full specification, none of the destination dummies has a significant coefficient, which is consistent with our main prediction that market integration contributed to equalizing art prices across the provinces of the Roman Empire.

TABLE 2 REGRESSIONS ON ROMAN STATUES (NOMINAL PRICES), I–III CENTURY AD

Notes: Robust standard errors are in the parentheses. Significance: *** p<0.01, ** p<0.05, * p<0.1.

Sources: Nominal prices from epigraphic sources.

TABLE 3 REGRESSIONS ON ROMAN STATUES (REAL PRICES), I–III CENTURY AD

Notes: Robust standard errors are in the parentheses. Significance: *** p<0.01, ** p<0.05, * p<0.1. Sources: Nominal prices from epigraphic sources. Adjustment for purchasing power from Wassink (Reference Wassink1991).

Finally, in Table 3, we confine our attention to the 123 commissions for which we have a precise date or a reliable estimate of the period in which they were commissioned and executed, and we adjust prices for the cost of living using the price index of Wassink (Reference Wassink1991); the observations are those displayed in Figure 2. The results show a negative but minimal trend in real prices, which is not statistically significant. The increasing and concave relationship between prices and the number of commissioned statues is now mostly significant, and the other main patterns of the earlier set of regressions are confirmed. In particular, the available evidence on the real prices of statues from the Roman Empire confirms that these prices were largely equalized across different provinces, supporting the hypothesis that increasing trade over a large integrated market was equalizing profit opportunities across destinations.