You can’t export that! Export ban for modern and contemporary Italian art

Abstract

Since 1939, an artwork in Italy can be subject to an “export veto” if it was created more than 50 years before the date of sale by an artist who is no longer living at the time of the sale. When the Italian bureau decides to exercise its right to veto exportation, these artworks cannot circulate outside the territory of Italy. Using original data from a hand-collected dataset covering all artworks made by non-living modern and contemporary Italian artists, auctioned at Christie’s and Sotheby’s in London and Milan between 2012 and 2016, we estimate a threshold model to consider the effect of the export veto law on price while controlling for the potential presence of a sample selection bias. We found that, while artwork prices are increasing in the time span between the year of creation and the date of sale, this effect reverses for artworks sold in Italy and created more than 50 years before the sale date. A similar pattern is also found in pre-sale estimates fixed by the auction houses, suggesting they exhibit rational behaviour in anticipating the export veto effect.

Appendices

Appendix 1: Export veto law and its links with art market issues

Export veto law has not only a potential direct impact on the art market through the veto risk as we explained above, but also a possible indirect link with other art market issues, that we are exploring in this section. While we could investigate the possible direct impact of the veto law on art market prices, our data do not allow us to check for these additional possible links. However, we think that presenting these issues can highlight the importance of our study, and possibly suggest future research on export limitations in the art market.

The (potential) link between export veto and home bias Export veto laws could in principle couple with (or be motivated by) a home bias effect, namely the fact that collectors of a given nationality will value relatively more the artworks by artists with the same nationality than would collectors of other nationalities. However, while the export veto law effect can be expected to be observed only when all the conditions defined by the law are met, home bias simply needs the nationality of the artist to be the same as the collector. The presence of home bias in the art market has been studied within the economics literature, starting with the study by Renneboog & Spaenjers (2011) on modern Russian art, where the authors investigate the relationship between the local and global stock market and the prices of the Russian artworks, explaining the dynamics they observe as a combination between a home bias effect and an investment choice made by the availability of greater wealth Russia after the mid-1990s. Castellani et al. (2012) study the heterogeneity of artist price using data from the Italian art market and find that an Italian artist traded in Italy presents a lower price heterogeneity than a non-Italian one, suggesting that this result can be due to a home bias effect. Later, Steiner et al. (2013) analyse how barriers to art trade in a certain country influence the composition of the collections of that country’s collectors, that is, they examine if home bias exists in art collections. They hypothesize that the tougher the trade restrictions in a country, the higher the home bias of its collectors, namely that these collectors hold a higher proportion of pieces by artists from their own country. However, since openness is not computed explicitly considering laws that restrict trade, but as a ratio between the sum of exports and imports divided by the GDP of the country, which is a measure that could be also influenced by the demand itself, a caveat should be made when considering that their hypothesis is supported by the data.²² A further analysis has been developed by Renneboog & Spaenjers (2015), who test whether prices and returns in the international art auction market are influenced by geographical segmentation, considering both law-related barriers to trade (as the Italian case) and demand-related effects (that may be due to cultural preferences towards artists with the same nationality of the buyer). They find a lower effect by local factors for high-tier art, where an artist’s quality is proxied by the length of his/her biography in the online encyclopedia Oxford Art Online. Local factors are however important for other segments of the market. For example, the Italian deviation from the global trend could be explained by the presence of trade barriers, while the Australian deviation could be related to high transport costs (Karataç 2019). This result is confirmed by Vosilov (2015) for the sculpture market finding that the average price is higher for sculpture sold in the home country of the artist compared to outside it, and this effect is stronger for the low-tier segment of the sculpture market than for the high-tier segment. The author attributes this home bias effect to familiarity and patriotism, with the latter as a more persistent source of home bias than the former.

Export veto and its possible reflections on Artist Resale Right Besides its (potential) relationship with home bias, the export barrier introduced by law could, in principle, collide with other national laws, such as the Artist Resale Right (ARR) law. In the countries where this law exists, an artist has to be paid a royalty for each trade (after the first) of his/her artworks within the national territory if certain conditions are met. Historically, this type of legislation has existed in Italy, France, Belgium, and Czechoslovakia since the first half of the XX century. In 2001, a Directive on the resale rights of artworks was implemented by the European Union (2001/84/EC), suggesting how to develop national laws on the topic for the EU Member States. Other countries, such as the USA, Canada, and China, do not have an ARR law at the national level (for example, California has an ARR law), but some galleries autonomously pay resale rights to their artists (Boicova-Wynants, 2019).²³ For example, in Italy the ARR is paid to a living artist if one of his/her artworks is sold for at least 3,000 EUR by a professional, or to the artist’s heirs if he/she has been dead less than 70 years on the date of sale and the piece is sold for at least 3,000 EUR by a professional (Candela & Scorcu, 2010). It is easy to see that some artwork could meet the conditions for both the ARR and the export veto law, specifically those made more than 50 years before the sale date by an artist who died less than 70 years before the sale date, and were sold for at least 3,000 EUR. If the market shrinks due to the export veto effect, the heirs who would collect ARR could see their ARR decrease because a smaller market is likely to yield lower prices, and hence lower ARRs. Hence, knowing if the export veto law could have an impact on prices is important to understanding whether it would be necessary to emend the export veto law or the ARR law, so that they are more in line with each other.

Transactions costs and art trade By the traditional trade theory, transaction costs are the main cause of market imperfection. In the art market, the transaction costs such as insurance, transport, and storage costs are not negligible. At the auction, for instance, buyers and sellers must pay a fee on the sales price to the auction house. Collectors, galleries, and museums must bear costs to protect and preserve their artworks. Often, these costs are related to the size and material of the artworks, and they positively affect their price. On the other hand, transaction costs such as the distance between the trading partners hurt bilateral trade for cultural goods. At the same time, the GNP of the exporting country has a positive impact on cultural trade.²⁴ Also, cultural proximity, linked to language, ethnic background, religion, belief, trust, and opinions, positively affect bilateral trade. The main explanation for this positive effect is the reduction of trade costs induced by cultural proximity.²⁵ Not all these transaction costs must be supported when an agent wants to smuggle or illicitly export an artwork, avoiding the limitations of laws such as the Italian export veto. In this case, specific characteristics are more likely to be important than others, among which the provenance record (the more precise the record, the harder it would be to bring the piece out of the country, as also reported in Watson (1997)), the size of the artwork (the smaller the artwork, the easier would be to export it without notice, like what Oosterlinck (2017) showed for artworks movement during the WWII), the corruption of exporting country (Fisman & Wei, 2009), and the importance of the artist name, which is likely to attract the attention of public bodies that control the export.

Appendix 2: Is an artwork’s performance influenced by the applicability of the Veto law?

In this appendix, we propose a measure of performance of the auction in terms of sale (un)success and study if it is related to the application of the veto law. To build our index, we follow the approach used by the Monitor Aste research group of the Arteconomy - Il Sole 24 Ore to measure the auction performance. The Arteconomy research group uses an indicator with multiple levels: the lower level pools together unsold pieces and artworks with a hammer price below the minimum presale estimates, a second level has only pieces that sold at a price equal to the minimum presale estimate, a third level has only pieces that sold at a price equal to the maximum presale estimate, while the highest level is made of artworks with a hammer price higher than the maximum presale estimates. The indicator is used to signal an unsatisfactory result (lower level), a satisfactory result (third level), and very good performance (highest level). Notice that the index of auction performance can be calculated for an artwork or a group of artworks, for an artist by aggregation of her/his artworks or to assess the overall outcome of an auction or auction session. In our case, we will focus on the performance of each artwork.

Our auction performance measure is based on the difference between the hammer price and presale estimates. In particular, we have underperformance when the hammer price is under the minimum presale estimate (or the artwork is unsold), while we have overperformance when the hammer price is over the maximum presale estimate. This approach is useful to analyse whether the veto issue affects auction performance through the model (4) and (5) since it allows us to check if the bidders behave differently when they have different expectations about the veto law application, knowing there is a possibility that export of an artwork could be banned by law. The performance measure we use is an ordered categorical variable that we call $\texttt{Performance}$, equal to 0 when the piece is unsold or when the hammer price is below the minimum presale estimate, equal to 2 when the hammer price is over the maximum presale estimate, and equal to 1 otherwise. Table 5 reports the descriptive statistics for the variable $\texttt{Performance}$.²⁶

Table 5 Descriptive statistics of $\texttt{Performance}$. Values are rounded to the second digit

To check for the relationship between the performance and the applicability of veto, we propose a linear model for the probability of an artwork to be traded at an auction price under the minimum estimate, over the maximum estimate or between the minimum and the maximum estimate:²⁷

$$\begin{aligned} \mathtt{Performance}^*_{ijm}= & {} \gamma _1 \mathtt{Veto}_i + \gamma _2 \mathtt{Age}_i + \gamma _3 (\mathtt{Veto}_i \times \mathtt{Age}_i) + \gamma _4 \mathtt{Year}_i \nonumber \\{} & {} \quad + \gamma _5 \mathtt{Year}_i^2 + A_j+ M_m + \epsilon _i \end{aligned}$$

(B.1)

where the covariates are the same in (5) with the addition of $\texttt{Year}^2$, $\epsilon _i$ is an error component and the $\gamma _k$ are the usual parameters to be estimated. Since $\texttt{Performance}^*$ is an unobservable latent variable, what we can observe is:

$$\begin{aligned}{} & {} \texttt{Performance}=0 \;\;\textit{if} \;\;\;\;\;\;\;\;\;\;\;\; \texttt{Performance}^*\le \mu _0 \nonumber \\{} & {} \quad \texttt{Performance}=1 \;\;\textit{if} \;\;\;\;\mu _0<\texttt{Performance}^*\le \mu _1\nonumber \\{} & {} \quad \texttt{Performance}=2 \;\;\textit{if} \;\; \;\;\mu _1<\texttt{Performance}^*\end{aligned}$$

(B.2)

where $\mu _0$ and $\mu _1$ are the cut-points. In particular, we perform an ordered logit model where the $\texttt{Performance}^*$ is estimated as a linear function of our independent variables and our cut-points. The probability of observing our response variable corresponds to the probability that the estimated linear function, with a random error, is within the range of the cut-points estimated for the response variable.

To check for potential differences between the model for the pieces traded in Italy and those traded abroad, we estimated (B.1) in the two subsamples defined by $\texttt{Italy}$. The results of the estimation are reported in Table 6.

Table 6 Estimated odd ratios of the model in (B.1)

The results reported in Table 6 are in line with what we found in Sect. 4.1, suggesting that the veto applicability is taken into account by collectors when they bid on pieces by Italian artists sold in Italy.