Magoosh GRE

Gold prices recession: A critique of Starr & Tran (2008)

| March 14, 2015

1. Introduction / Summary
The following presents a critique of a paper by Starr and Tran (2008), ‘Determinants of the Physical Demand for Gold: Evidence from Panel Data’, which, as the title suggests, looks at the factors which impact upon demand for gold in its physical form: “jewellery, bars, coins and medallions” (Starr and Tran 2008, p. 416). This market has been boosted by demand from countries from the “emerging market”, I including India, China and the Middle Eastern countries, due to its portability, ability to hold value, and attractiveness to markets where the full range of investment opportunities has not developed. It also has scope for rapid liquidity and high cultural value.
The authors present an approach which differs to previous ones, which tend to look at “high-frequency data” on prices, rather than physical exchanges. Previous studies of this nature include Capie et al (2005) (CITE) and Cai et al (2001). There has been little research looking at emerging economies, and no previous study has looked at the ways in which social and cultural factors impact on demand for gold. It is also not clear whether previous studies, which seem to link volatility in exchange or inflation rate with increased demand for gold, can be extended to map the demand for physical gold.
In order to address the gap in research, Starr and Tran interrogate panel data from the World Gold Council covering 21 countries over a 22 year period to 2003, assessing importance of factors including price, income levels, volatility of inflation and exchange rate and credit access upon consumer demand for jewellery and retail investment (coins and bars). The use a formula to calculate demand for gold based upon these variable factors.
The authors identify problems in estimating the equations, including potential bias, and the ways they addressed these issues. For example, the problem of bias was corrected by Bruno’s (2005) method of running the panel data model with “a correction for the bias associated with the lagged dependent variable”

The authors present their results. They ran five different specifications: Pooled OLS Fixed Effects Bias-Corrected LSDV GMM Difference GMM System, all of which “show a considerable amount of persistence in physical demand for gold”, meaning that if demand for gold falls, it takes some time to return to its previous level, which is consistent with gold’s popularity being in part due to what other people in the country think about gold. Their results also suggest that GDP is not significant in determining demand for gold, nor is per capita income, but that demand is higher when income is volatile or has not grown recently. They suggest that this means that gold’s value rises in economically unsettled times. They also find that high stock market growth leads to demand for physical gold. Finally, factors which have been found to drive demand for gold in other studies have no relationship with physical demand. They conclude that the factors driving demand for physical gold are quite different from those driving demand for gold in other forms.

Results are also split by developed and developing countries. Here, lagged demand’s effects are lower in the developed than in the developing world (strong economies are better able to withstand shocks to gold demand); per-capita income effects gold demand only in developed countries, both sub-samples seem to purchase gold in times of uncertainty; inflation drives demand up in developed countries, and demand is higher where private credit markets less developed.

Their results, they suggest, show not only that different factors impact upon demand for physical gold and gold claims, and that there are differences between demand in the developed and developing world, but also that “there are indeed persistent unobserved heterogeneities in gold demand, consistent with a role of socio-cultural factors”, with such differences accounting for 68% of gold demand variation. This, they suggest, is likely to be due to social and cultural factors, pointing out that demand is high in countries where it is valued culturally. They further suggest that the impact of social and cultural factors might explain why “policy measures intended to discourage investment in physical gold tend to be ineffective”
2. Critique
Overall, this is an interesting and persuasive paper, and one which seems to investigate new ground as well as highlight areas for further investigation. This section will look in more detail at particular aspects of the paper. In terms of structure and style, the paper is clearly written, and complex economic data is balanced with clear, well written explanations. However, there is no overall perspective at the start of the paper giving an overview of what will be covered, and a summary of key findings. This would normally be found in an abstract, or as part of the introduction, and would improve the paper. The structure of the paper is clear, and they follow an accepted format by discussing the nature of the problem and current literature, explaining their methodology, presenting results, and discussing their findings. Their arguments are easy to follow and clear.
The introduction is clear, and sets out the context for the study. There could perhaps be more discussion of previous work in the field. While the authors do suggest that there is only a limited amount of work, the references are perhaps a little limited. There could, perhaps, also be more detailed critique of other studies. Most mentions of other studies simply summarise what they find, rather than critiquing their methodology, sampling techniques and related areas. For example, the results of a study by Haugom are discussed, and similarities to their own study pointed out, but it might have been interesting to look in more detail at how the two studies were similar, how they differed, and whether Haugom’s had any problems.
In terms of methodology, this section of the paper seems very good. The methods used and the nature of the data are clearly explained, particularly the nature of the regression and the variables upon which it is based. The justification of inclusion of the variables is also useful. Explanations, for example of the four categories of gold, are clear. They also identify econometric problems that were raised by the equation used, and the discussion of how this was resolved is good.
A couple of questions could be asked, however. First, there is no justification of the data selection. It would have been useful to have explanation of why data from the World Gold Council was chosen, whether there were other viable alternative sources, and the reliability of their data. In addition, the methods seem less than adequate to address one aspect of the problem they set for themselves, that is, to measure to what extent social and cultural factors influence demand for physical gold. This might be a function of the lack of useful data within the WGC data set, however.

In terms of the author’s discussion and conclusions, these are well presented and seem to follow from their results. They make excellent use of tables to summarise and present information in a way that makes it easier to digest. They not merely interpret the data in terms of physical gold demand, highlighting for example the persistence in demand, but also make suggestions about the further interpretations which can be made, for example suggesting that gold demand is higher when income is more volatile because physical gold is more highly valued at times of economic uncertainty. The discussion of differences between emerging and developed countries and their demand for physical gold is also useful, and related back to topics raised in the introduction well.
However, the link the authors seem to claim between high demand for physical gold and social or cultural value seems less well argued from the evidence. Their discussions rely upon “uncontrolled, unobserved heterogeneity”, with this accounting for 68% of gold demand variation, rather than linking differing demand to specific socio-cultural variables. While they acknowledge that “socio-cultural factors may be only one such difference” they go on to claim “they seem likely to make an important contribution”, without giving robust evidence why this is the case. The author’s point would be more convincing if illustrated with hard research data suggesting that gold is valued more highly than others in certain countries.
While their conclusion section is clear, and summarises their discussion well, it would have been good to have some recommendations for future research, for example more detailed research on the cultural value ascribed to gold in different countries. They also restate that “culture seems quite likely to be involved” as a determinant of demand for physical gold, without providing evidence why this is likely.
3. Conclusion
This has presented a short critique of a paper by Starr and Tran (2008), which seeks to use panel data to investigate factors influencing the physical demand for gold. The paper was summarised, and key aspects of the literature review (introduction), methodology, results and discussion were critiqued. Overall, this is a very interesting paper which highlights the differences between developed and emerging nations in terms of demand for gold, and which suggests that differences between countries might be due to social or economic factors.


Bruno, G S F (2005), ‘Approximating the Bias of the LSDV Estimator for Dynamic Unbalanced Panel Data Models’, Economics Letters, 87, 361–66.

Cai, J Y-L C and Wong, M (2001) ‘What Moves the Gold Market?’, Journal of Futures
Markets, 21:3, 257–78.

Capie, F, Mills, T C and Wood, G (2005) ‘Gold as a Hedge against the Dollar’, Journal of International Financial Markets, Institutions and Money, 15:4, 343–52.

Starr, M and Tran, K (2008), ‘Determinants of the Physical Demand for Gold: Evidence from Panel Data’, The World Economy, 416-436.

Category: Essay & Dissertation Samples, Finance Essay Examples