Exploration externalities and government subsidies: The return to government

Resources Policy 47 (2016) 78–86

Contents lists available at ScienceDirect

Resources Policy journal homepage: www.elsevier.com/locate/resourpol

Exploration externalities and government subsidies: The return to government James J. Fogarty a,n, Simon Sagerer b a b

School of Agricultural and Resource Economics, University of Western Australia, 35 Stirling Highway Crawley, 6009 Western Australia, Australia ACIL Allen Consulting, 28 The Esplanade, Perth 6000, Western Australia, Australia

art ic l e i nf o

a b s t r a c t

Article history: Received 8 September 2015 Received in revised form 28 December 2015 Accepted 4 January 2016

Governments have, for a long time, invested in the direct provision of basic geological survey information to support exploration and mining activity. Recently, Australian governments have also started to provide direct drilling subsidies to exploration companies. Using data for Western Australia we investigate the return to government from the direct provision of geological survey information and the provision of drilling subsides. We find no evidence that drilling subsidies are less effective than traditional geological survey spending in generating a return to government. We suggest drilling subsidies are effective because there is a dishonesty externality in the market for exploration equity capital that gives rise to a market for lemons problem, and that government programs to award drilling subsidies to exploration companies work as a third party certification system that addresses this problem. We conclude by showing that, with real discount rates of 5%, 7%, and 9%, and a narrow definition of benefits, the expected benefit–cost ratios for State government support for exploration are 9.0, 6.7, and 5.2. & 2016 Elsevier Ltd. All rights reserved.

Keywords: Exploration externalities Government policy Exhaustible resources JEL: Q32 Q38

1. Introduction In this paper we find that, since 1990, for every $1M spent by the Western Australian government on support for the exploration sector there has been additional new private sector exploration spending of at least $5.5M. We also find that direct drilling subsidies are at least as effective as traditional geological survey spending in stimulating new private sector exploration activity. We interpret this finding as evidence government drilling subsidies do more than just lower the marginal cost of drilling at an individual firm. Specifically, we suggest that because exploration prospects can be marketed in equity capital markets honestly or dishonestly there is a market for lemons problem in the exploration equity capital market, and that the award of a government drilling subsidy works as a kind of third party quality assurance mechanism to mitigate this market for lemons problem. We conclude by showing that, with real discount rates of 5%, 7%, and 9%, and a narrow definition of benefits, the expected benefit–cost ratios for State government support for exploration are 9.0, 6.7, and 5.2. 1.1. Background The mining industry is one of Australia's most important n

Corresponding author. E-mail addresses: [email protected] (J.J. Fogarty), [email protected] (S. Sagerer). http://dx.doi.org/10.1016/j.resourpol.2016.01.002 0301-4207/& 2016 Elsevier Ltd. All rights reserved.

industries, and is especially important in Western Australia. For example, in the 2013 financial year the mining industry contributed 29% ($71B) of Western Australia's Gross State Product (GSP) (Government of Western Australian, 2014a, p. 2). Mining royalty income for the Government of Western Australia is also significant, contributing around $6B, or 22% of general government revenue (Government of Western Australian, 2014b, p. 88). Exploration activity is a necessary precursor to mining, and in Australia, both State and Territory governments provide substantial subsidies to support the exploration sector. For example, in addition to the annual geological survey budget allocation of around $25M, over the period 2009–2017 the Government of Western Australia has committed $130M in spending to support greenfield exploration, including the provision of drilling subsidies.1 Western Australia is not alone in increasing the level of government support for the exploration sector, and a review of State and Territory websites found widespread evidence of new government programs to support exploration activity.2 The Commonwealth government, through Geoscience 1 Exploration Incentive Scheme: Available www.dmp.wa.gov.au/7743.aspx (accessed 13.05.15). 2 Northern Territory: Available www.core.nt.gov.au/about.html; South Australia: Available www.pir.sa.gov.au/minerals/initiatives/pace; Queensland: Available www.dnrm.qld.gov.au/our-department/policies-initiatives/mining-resources/ future-resources-program; New South Wales: Available www.resourcesandenergy. nsw.gov.au/miners-and-explorers/geoscience-information/about/new-frontiers (accessed 05.12.14).

J.J. Fogarty, S. Sagerer / Resources Policy 47 (2016) 78–86

Australia, also has a commitment to the provision of a broad range of geoscience information services. The economic literature concerning exploration and extraction is substantial, with Cairns, (1990) and Krautkraemer, (1998) providing reviews of relevant issues. Here, the intention is not to provide a complete review of the literature, but to focus on why governments may choose to provide financial subsidies to support mineral exploration, and to estimate the return to government from the provision of exploration subsidies. Exploration activity is associated with an information spillover. The success or failure of a drilling campaign provides important information about the success or failure of other drilling campaigns in similar regions; where similarity could be defined in terms of geographic distance or mineralisation. Private sector firms involved in primary greenfield exploration work are therefore not able to capture all of the benefits of their investment. Early evidence provided in Peterson, (1975) illustrates that the information spillover effect is real. To address this issue Peterson concludes with a series of public policy recommendations, including the direct government provision of geophysical studies, or subsidies to the private sector to undertake these studies. Subsequent work, for example Dodds and Bishop (1983) also suggests a role for public information provision. Public geological survey services have a good reputation. For example, in the US, going back to at least the 1850s, the provision of reliable government survey work has been identified as a critical element in the successful and rational development of resources that, in turn, led to the rise of the US as a global power (David and Wright, 1997, p. 227). In an Australian context, the general consensus in the literature is that Australian geoscience expenditure has more than paid for itself (Hogan, 2003; Productivity Commission, 2013). The information spillover externality is not, however, the only externality issue in the exploration market. Junior exploration companies generally raise funds from equity markets, and the promoters of a capital raising for an exploration company know more about the quality of the geological prospects than the providers of equity capital. For an investor, even ex-post, it is difficult to know if the failure of a given drilling campaign was due to the poor quality of the original drilling plan or not. Combined, these circumstances allow for any given exploration company capital raising to be marketed in either an honest or dishonest manner. If we accept that preparing a high quality portfolio of exploration prospects is more expensive than preparing a low quality portfolio of exploration prospects, then the circumstances for a market of lemons to arise are met. In Akerlof's model this dishonesty effect drives the size of the market to zero, and the legitimate businesses that are driven out of existence are characterised as an externality cost (Akerlof, 1970, p. 495–6). One reason there is not a complete collapse in the market for exploration funds is the existence of counteracting institutions. For example, the reputation of individual explorers is one mechanism that works to mitigate against a total collapse in the market for exploration equity capital; but there is a limit to the role individual explorer reputation can play. Company branding can also mitigate against the development of a pure market for lemons. For example, if the ABC Nickel Exploration company has been successful, then company management could launch the ABC Gold Exploration company, and through this type of ‘chain’ company branding signal that the same successful management practices used at ABC Nickel Exploration will be applied at ABC Gold Exploration. The role of the stock broker as a specialist advisor on prospect quality also helps to mitigate the development of a market for lemons, but the incentives for the broker are not the same as for the investor; and even if broker and investor incentives could be aligned, because people can free ride on broker research notes there would

79

still be underinvestment in the provision of quality assurance information. Counteracting institutions do not, therefore, completely resolve the market for lemons problem. Although the dishonesty cost disappears with vertical integration of the exploration function with the mine operation function, in the mineral sector the trend has been for large mining companies to outsource exploration. Outsourcing exploration is cost effective for large mining companies as for any given exploration program the health, safety, internal compliance costs, and labour costs at large mining companies are much greater than at junior exploration companies. Outsourcing exploration is also a safe business option for large miners as for any discovery significant enough to be of interest to them, a junior exploration company will not be able to raise the funds required to develop the project. This in turn allows large mining companies to enter the process at the post-greenfield exploration pre-mine development stage. As such, large mining companies retain access to significant new development sites even though the exploration function has been outsourced. Here we estimate the return to the Government of Western Australia of spending to support greenfield exploration and augment our empirical modelling results with information found in ASX listed exploration company announcements that provide support for our empirical results.

2. Estimating the exploration response If we think about the exploration process in generally, when commodity prices or government policy change it is unlikely firms will respond immediately. For example, it takes time to research the most prospective sites. It takes time to obtain management and board level approval for a specific exploration program. It takes time to source the equipment needed for a drilling program and then get the equipment on site, etc. Given this characterisation of the market, there is then a very real problem in assessing the impact of price and government policy changes. What is required for policy evaluation is a measure of the long-run exploration expenditure response; what is observed every period is the short-run exploration expenditure response. Here, to estimate the long run private sector response to government spending we use the Autoregressive Distributed Lag modelling approach of Pesaran and Shin (1999). 2.1. Data The focus of the research is an evaluation of State government expenditure to support exploration in Western Australia. However, we also have details on funding allocated specifically to drilling subsidies, as well as the traditional geological survey allocation, and so we also estimate a model where these two expenditure types are considered separately. The drilling subsidy program is quite recent (see Fig. 1) and so it is necessary to be cautious when interpreting the results, but these initial results are still informative. The expenditure information has been provided by the Department of Mines and Petroleum, Government of Western Australia. SNL Metals & Mining provides information on global exploration, development, and production activity. Based on information in SNL Metals & Mining (2014a), the most common deposits in Western Australia are gold, nickel, and iron ore; and over the last 35 years 96% of the discoveries with commercial potential were associated with one of these three minerals. We therefore include the Australian dollar price of these three commodities in our model. For consistency, both the price series and exchange rate values have been sourced from the London Metal Exchange.

80

J.J. Fogarty, S. Sagerer / Resources Policy 47 (2016) 78–86

Fig. 1. Summary of key data series Source: London Metal Exchange; Australian Bureau of Statistics; Government of Western Australia.

Independent of market conditions, we think discoveries could trigger an increase in exploration activity. We investigate this potential effect through the use of a variable created using the details in SNL Metals & Mining, (2014a). Specifically, we count as a discovery all new discoveries in the database, and any increase in the resource if there has been no change in the resource for at least five years. We include this variable in the model as a count variable. The dependent variable in the model is private sector exploration spending in Western Australia, and the exploration data is published in ABS (2014a). The data frequency is limited by the government reporting period, which is annual. The data series runs from 1989 to 2014. 2.2. Modelling approach The approach we rely on for modelling the private sector response to government spending can be understood as follows. Let: yt denote exploration expenditure at time t; p1t , p2t , p3t denote, respectively, the Australian dollar price of gold, iron ore, and nickel at time t; Gt denote government expenditure to support exploration at time t; Dt denote a count measure of exploration discoveries at time t; Tt denote time; and Δ denote the change in a variable such that the general form of the model estimated can be written as:

Δyt = α + ∑i βi Δyt − i + ∑j φj Δp1t − j + ∑k γk Δp2t − k + ∑l ηl Δp3t − l + ∑m λm ΔGt − l + ∑n τn Ttn + δDt − 1 + φ0 yt − 1 + φ1p1t − 1 + φ2 p2t − 1 + φ3 p3t − 1 + μGt − 1 + et ,

where the Greek letters denote parameters to be estimated and et is a zero mean constant variance stationary error term. In words, the model says that the change in exploration expenditure observed today can be explained by: (i) the historical commodity price level and changes in commodity prices; (ii) the level of past government investment to support exploration and changes in the level of government investment to support exploration; (iii) the change in exploration expenditure in previous periods; (iv) new discoveries in the previous period; and (v) an autonomous time trend. Following Pesaran and Shin (1999), the specific steps in the modelling approach where: (i) establish that the order of integration for all data series is less than two I (2); (ii) estimate an unconstrained Error Correction Model (ECM) of the form shown above, with model selection guided by AIC, the statistical significance of individual coefficients, and general regression model specification tests; (iii) check the dynamic stability of the model; (iv) conduct the bounds test to establish whether or not there is a long run relationship; and (v) derive the long run response and associated standard error. 2.3. Modelling results Summary information on the Kwiatkowski–Phillips–Schmidt– Shin tests for stationary are shown in Table 1, and based on the test results we conclude that there is no evidence any series is I (2) in levels. The results for the unconstrained ECM model are shown in Table 2. Although many of the coefficients are statistically significant at conventional levels, there are multicolliniarity issues

J.J. Fogarty, S. Sagerer / Resources Policy 47 (2016) 78–86

81

Table 1 Summary stationary test results (p-values). Test format

Levels (no trend) Levels (trend) 1st Difference (no trend)

Price series

Expenditure series

Iron ore

Nickel

Gold

Exploration

Government

Traditional

Co-drilling

o 0.01 o 0.01 4 0.1

0.03 40.1 40.1

o 0.01 o 0.01 40.1

o 0.01 o 0.01 40.1

o0.01 0.02 40.1

o 0.01 0.02 4 0.1

o0.01 o0.01 0.09

Note: The test output reports only 4 0.1 when true p-value is greater than this.

with the model. Specifically, for five explanatory variables the Variance Inflation Factor is greater than 100. As the model version allowing for discovery effects fails both the Breusch–Godfrey autocorrelation test and the bounds test for establishing a long run relationship, to estimate the long run relationship we rely on the model results without this variable. As the model has only one lag of the dependent variable, the dynamic stability condition requires the absolute value of the lagged exploration expenditure coefficient to be less than one, which it is. The F-statistic for the bounds test is above the relevant critical value reported in Pesaran and Shin (1999), so we conclude there is a long run relationship. The long-run response is found as the lagged government spending coefficient divided by the negative of the lagged exploration coefficient, and the associated standard error can be found via the delta method. The relevant long-run response information is given Table 3. The first row of the table provides information on the long run private sector exploration response to government spending to support exploration, and the 95% confidence interval ranges from a modest positive response of $5.5 in new exploration spending for every $1 invested by the government to an implausibly large response of $254.4 in new exploration spending for every $1 of government spending. The lack of precision is due to the multicoliniarity problem. Based on this result, the conclusion we draw is that, in Western Australia, government spending to support exploration activity has stimulated a real increase in private sector exploration activity, but the true extent of the response is uncertain. Next, we investigate the impact of the recent direct drilling Table 2 Summary of regression model results. Est Intercept Exploration spending (L-1) Government spending (L-1) Iron ore price (L-1) Nickel price (L-1) Gold price (L-1) Δ Exploration (L-2) Δ Iron ore price Δ Nickel price Δ Gold price Δ Government spending Time Time squared Discoveries (L-1) AIC Breusch–Godfrey test (p-value) RESET (p-value) Bounds test (F-statistic)

SE ***

2241.0 0.522** 67.90*** 11.57*** 0.020*** 2.533*** 0.355 6.218* 0.005 1.372*** 41.52*** 116.40*** 8.83*** 277.70 0.10 0.20 12.15

(480.8) (0.206) (16.63) (4.58) (0.003) (0.671) (0.365) (2.927) (0.003) (0.350) (11.52) (30.18) (1.701)

Est 1710.0 0.316 46.34* 6.74 0.017 1.613 0.460 4.671 0.005* 1.09** 34.82** 97.34** 7.02*** 13.27 275.25 0.02 0.30 3.84

SE **

(598.8) (0.247) (22.25) (5.60) (0.004) (0.924) (0.358) (3.015) (0.003) (0.392) (12.04) (31.96) (2.086) (9.573)

Note: Log models were considered and did not perform well. Models with different lags on exploration were also considered but were strongly rejected on Akaike Information Criterion (AIC) based weight measures.

Table 3 Estimated long run response. Long run response

Est.

SE

Total spending Traditional Drilling subsidy

129.97 80.30 432.92

56.55nn 29.80nn 176.04nn

*** Significant at 1% levels. * Significant at 10% levels. **

Significant at 5% levels.

subsidy program. To do this we re-estimate the model for the full sample period, but separate out the drilling subsidy expenditure from the traditional geological survey funding allocation. For this model, after checking that the conditions for valid estimation of a long-run relationship still hold, we then derive the relevant longrun responses, which are shown in the last two rows of Table 3.3 Given the uncertainty surrounding these estimates, and given the short history of the drilling subsidy program, we do not want to over emphasise these findings, but at the moment, it does seem reasonable to conclude that, in terms of stimulating a private sector response, there is no evidence drilling subsides are less effective than traditional geoscience spending. 2.4. Qualitative support In light of the above findings it is reasonable to ask whether or not there is any real world evidence of government spending on geoscience and/or the provision of subsidies for drilling leading to new exploration activity. For evidence we started by looking at the announcements of companies listed on the Department of Mines and Petroleum website as having received a drilling subsidy. These companies are not a random sample of exploration companies, but our intent was to look for evidence, not consider a random sample of exploration companies. In terms of the impact of general geoscience expenditure, the history of Enterprise Uranium is a useful example. In December 2012 Enterprise Uranium Limited listed on the ASX after raising $5.1M in funds for grass roots uranium exploration. The exploration program of the company was detailed in the company prospectus, and the total new exploration expenditure approved for the first two years following listing was $1.8M if no new tenement applications were approved; and $3.7M if all new tenement applications were approved. The prospectus makes it clear that several of the exploration prospects that formed the basis of the assets of the company were identified as a direct result of historical and more recent work by the Geological Survey of Western Australia (GSWA). The uranium targets in the Perenjori area were identified by the

*

Significant at 1% level. Significant at 5% level *** Significant at 10% level **

3 The matching model estimated has a lag of two periods on exploration spending rather than one.

82

J.J. Fogarty, S. Sagerer / Resources Policy 47 (2016) 78–86

Company from airborne survey data recently flown by the Geological Survey of Western Australia (GSWA) under the Royalties for Regions program. Using this GSWA data and other proprietary data, Enterprise has assembled a strong portfolio of projects at Byro, Yalgoo, Peranbye, Ponton and Harris Lake, covering a total area of 5931 km2. These areas cover airborne uranium anomalies located over present day lake systems which have historically had little or no previous uranium exploration (Enterprise Uranium 20 December 2012). So, basic geoscience information allowed the subsequent identification of potential deposits, and companies really do form and raise funds for exploration based on information published by GSWA. Given the geoscience information is non-rival in consumption, the strong private sector response found as part of the quantitative analysis seems reasonable. Projects that receive a drilling subsidy from the Government of Western Australia are evaluated by an expert panel, and companies can use the award of a drilling subsidy as a quality signal in the market for equity capital. For example, the Antipa Minerals 9 December 2014 announcement of success in obtaining a drilling subsidy included the following passage: Antipa would like to acknowledge the ongoing support provided by the WA Government through its EIS [Exploration Incentive Scheme] programme for the Company's exploration programmes. Since listing the Company has successfully applied for four WA Government EIS co-funded drilling grants. The EIS co-funded drilling programme preferentially funds high quality, technical and economically based projects that promote new exploration concepts and are assessed by a panel on the basis of geoscientific and exploration targeting merit. The Sirius Resources Fraser Range Nickel project is another useful example. This prospect made use of data from various sources, including historical data from the 1960s, and GSWA soil sample information; and the company also received a drilling subsidy for its initial drilling program. From company announcements we can see that receipt of a drilling subsidy and association with a reputable prospector are things the company promoted when trying to attract investors. For example, in a presentation to the London Mines and Money Conference, December 2011, when the company had a market capitalisation of $13M, Sirius advertised that initial drilling at the Fraser range project had found nickel, copper, and cobalt enriched material and that “a “stratigraphic” diamond drilling hole (co-funded by the WA government exploration incentive scheme) is underway”; and that the company was “Backed by Mark Creasy-Australia's most successful prospector”. This behaviour is consistent with the idea that there is a market for lemons problem in the market for exploration equity capital.4 Sirius Resources has also moved to the development stage of the Fraser range project, and as described in the Annual General Meeting presentation, November 2014, the company market capitalisation was $1.2B. It is a very real possibility that the existence of a company such as Sirius Resources, that has received a State government drilling subsidy, increased in value by a factor of around 100, and quickly moved to the development stage of a medium sized nickel project, could reinforce views in the market place that receipt of a government drilling subsidy is a quality signal. Enterprise Metals provides another example. In June 2009 4

All presentation material available: www.siriusresources.com.au/investorpresentations.php (accessed 13.05.15).

Enterprise Metals had cash at bank of $1M. In the previous quarter exploration spending had been around $700,000, and in the forthcoming quarter planned exploration spending was also $700,000. The exploration budget of the company was therefore soon to be exhausted. In June 2009 the company announced that it was successful in obtaining a government drilling subsidy of up to $110,000. In July 2009 (not long after the collapse of Lehman brothers) the company raised $2.3M in equity capital for additional exploration work.5 That Enterprise Metals was able to raise additional funds for exploration just after announcing the receipt of a drilling subsidy does not mean that receipt of the subsidy was a causal factor in the successful capital raising. The example does, however, illustrate that there are genuine reasons to think receipt of a government subsidy for exploration drilling might allow firms to leverage this award to attract significant additional private sector funds for new exploration work. The original economic logic that supports the provision of direct drilling subsidies is that subsidies lower the marginal cost of exploration, and hence correct the underinvestment in greenfield exploration. The way companies actually promote receipt of a drilling subsidy suggests there may be a further mechanism at work. Specifically, if we can characterise the market for exploration equity capital as a market for lemons, and if receipt of a drilling subsidy is seen in the market as providing some sort of third party certification regarding the business management team and their drilling program, award of a drilling subsidy may allow the company to attract additional investment funds for exploration. Further, because the dishonesty externality drives the size of the market for exploration equity capital to zero if there are no counteracting institutions, the equity capital raised by any individual firm is new equity capital and not a transfer from other firms. 2.5. Comparative perspective In a review paper, Duke, (2010) reports estimates from 19 studies on the effectiveness of tax incentive and subsidy programs aimed at encouraging exploration, and across these studies the mean increase in private sector exploration per dollar of government investment is $6.2 (maximum $19). For Canada, Khindanova (2012, p.85) found that the introduction of the Minerals Exploration Depletion Allowance, which was in operation for the period 1983–97, was associated with a 62% increase in exploration activity by junior explorers, but no response from major mining companies. Earlier research by the Canadian Department of Finance also found that the deductibility introduced as part of Canada's Flow Through Share Scheme (similar to Australia's Franking credit system) altered the incentives for exploration such that for every $1 in lost tax revenue the incremental gain in mining exploration activity was $3 (Department of Finance, 1994). So, the upper bound of the response estimate found in this study far in excess of what has previously been reported in the literature, but the lower bound of the response estimate is consistent with what others have found.

3. Return to government Ultimately, governments provide exploration incentives because they hope to see the development of new mines. Gold, 5 All ASX announcement material and company presentations are available at http://enterprisemetals.com.au/investor-information/asx-announcements/ (accessed 10.12.14).

J.J. Fogarty, S. Sagerer / Resources Policy 47 (2016) 78–86

nickel, and iron ore mines dominate new mine discoveries in Western Australia, and as such we limit our evaluation to expected future small, medium, and large mines for these three commodities. Much of the return from mining is in terms of company tax receipts and income tax paid by mining employees. However, as this revenue flows to the Commonwealth government, and here the focus is on the return to the government of Western Australia, these tax receipts are not counted. To estimate the expected return to the State government we explicitly consider the role of uncertainty in future commodity prices and the private sector response to government spending. The specific process we use to estimate the expected benefit–cost ratio involves the following steps. Step one: draw the expected exploration multiplier value from an assumed distribution; Step two: use this multiplier value to calculate the number of expected discoveries of each type due to $1M in government spending; Step three: draw the relevant commodity price values from the assumed distribution; Step four: determine, from the exploration company point of view, whether or not the NPV of transitioning each discovery to an operating mine is positive; Step five: for all potential new mines with a positive NPV from the exploration company point of view, calculate the return to government from payroll and royalty taxes for a given discount rate; Step six: repeat 10,000 times; Step seven: vary the government discount rate assumption and repeat a further 10,000 times. For each discount rate this process generates a distribution of expected benefit–cost ratios. We then plot this distribution to calculate the: (i) mean benefit–cost ratio; (ii) median benefit–cost ratio; and (iii) proportion of the benefit–cost ratio distribution greater than one. How we develop the relevant values to support the modelling is explained below. 3.1. Calibration values To determine per metre drilling costs we rely on information in ABS (2014a) which indicates that the three year average cost of exploration drilling was $360 per metre, in real 2014 dollars. To determine discovery rates we rely on information in SNL Metals & Mining (2014b). For the period 1980–2013 the SNL database reports the (annual) reserve and production history of 1196 named mineral deposits in Western Australia. We define a new discovery as an increase in the reported reserve if the reserve has not increased in the previous five years. The total discovered quantity was estimated by adding any subsequently discovered reserves to the initial discovery. Across this period we find that 314 discoveries were made in Western Australia. Using the reserve criteria shown in Table 4 to define deposits with commercial potential implies 116 of the 314 discoveries were large enough to potentially become a mine. We only have access to WA specific drilling and discovery rates for 2013 and 2014 (SNL Metals & Mining, 2014a). In these two years 2.7 million metres were drilled and 37 discoveries were made. If we assume that the ratio of commercially viable discoveries is the same as the long run average (116/314) this translates into 14 commercially viable discoveries, or a commercially viable discovery for every 2.3 km drilled. This is a higher discovery rate than reported in Schodde (2014, p. 23) for Australia as a whole, but given Western Australia is the dominant location for major discoveries in Australia (Geoscience Australia, 2014, p. 3, 10– 11); and given small gold mines are a feature of mining in Western Australia, it seems appropriate that the discovery rate in Western Australia should be above the national average. We then apply a reserve criteria to define small, medium, and large mines to work out the relationship between total metres drilled and a discovery of any given mine type. Table 4 shows the reserve thresholds we apply for each mine type, the number of

83

Table 4 Stylised deposits: characteristics. Characteristic

Units

Gold

Iron ore

Nickel

Minimum reserve small Minimum reserve medium Minimum reserve large Number of small discoveries Number of medium discoveries Number of large discoveries Drilling per small discovery Drilling per medium discovery Drilling per large discovery

Ton (M) Ton (M) Ton (M) No. No. No. Metre (M) Metre (M) Metre (M)

4 30 100 53 15 4 0.52 1.83 6.87

100 800 1500 13 7 4 2.11 3.93 6.87

2.5 15 100 14 3 3 1.96 9.16 9.16

commercially viable deposits of each mine discovered over the period 1980–2013, and the expected number of metres that need to be drilled to find a discovery of each type. Size thresholds were developed with reference to actual operations in Western Australia, as reflected in SNL Metals & Mining (2014b). For example, the smallest gold mines currently operating in Western Australian produce about two tonnes of gold per annum. We find that the average mineral grade for this type of mine is 3.0 g/t; so assuming a minimum required mine life of five years, the smallest reserve size with commercial potential is four million tonnes. Medium gold mines such as Agnew or Sunrise Dam produce approximately seven tonnes of gold per annum. Assuming a minimum life of 10 years for this type of mine, and an average reserve grade of 2.4 g/t, the minimum reserve size of a medium mine is 30 million tonnes. Using similar logic it was possible to determine a threshold classification for all nine stylised deposits. The production assumptions for each of the representative mines are based on information in the SNL Metals & Mining (2014c) for benchmark mines, and are summarised in Table 5. In the case of gold mines the small and medium mines are assumed to be underground mines, but the large gold mine is assumed to be an open cut mine. For the nickel projects, both the small and medium mines are assumed to be sulphide nickel projects. The large nickel mine is assumed to be a nickel laterite project. The chemical process of separating the mineral from the rock is less complex for nickel sulphide mines and this is reflected in the operating expense assumptions. All iron ore mines are assumed to be hematite mines. For all mines we assume an average FTE wage of $132,000, which is based on ABS (2015). The construction assumptions use the same set of benchmark mines as the production assumptions, but the data is sourced from company reports and the REPS Major WA Projects List. Construction assumptions specify capital expenditure, the time required from exploration campaign to mine construction, construction time, and production ramp up profile. As discussed above, multicolinearity issues mean it is not possible to determine precisely the private sector response to government spending that supports exploration. Using the results reported in Duke (2010) as a reference point, we think a conservative approach is to use only values between the 95% and 90% lower bound confidence interval from our model. Specifically, we assume the exploration response is drawn from a uniform distribution bounded by 5.5 and 28.4. As shown in Fig. 1, there is significant price volatility for each commodity. Given changes in operating costs as mines have expanded in recent years, there is genuine uncertainty regarding what might represent reasonable price scenarios. Here we assume the average price that applies over the life of the mine is drawn from a uniform distribution where the distribution bounds are $700 per ounce to $1400 per ounce for gold, $12,000 per tonne to $24,000 per tonne for nickel, and $55 per tonne to $110 per tonne for iron ore.

84

J.J. Fogarty, S. Sagerer / Resources Policy 47 (2016) 78–86

Table 5 Key production information on stylised mines. Source: Author analysis of SNL Metals & Mining Databases; Company reports; Chamber of Commerce & Industry Western Australia (2013). Mine detail

Units

Small

Medium

Large

Gold Mines Capital Expenditure Discovery to production lag Construction time Ramp up time Operating expenditure Employment Average grade Annual Production Resource

$AsUD (M) Years Years Years $AUD per oz FTE grams per tonne MT MT

80 2 0.5 0.5 850 250 3.0 1 9

800 5 1.5 0.5 1000 500 2.4 5 52

3500 10 2.5 1 1120 750 1.6 30 429

Nickel Mines Capital Expenditure Discovery to production lag Construction time Ramp up time Operating expenditure Employment Average grade Annual Production Resource

$AUD (M) Years Years Years $AUD per tonne FTE % Ni MT MT

80 2 1 0.5 10,000 100 1.70 0.5 8

500 5 1.5 1 8000 225 1.30 1.5 25

2000 10 4 2 18,000 390 0.80 3 171

Iron Ore Mines Capital Expenditure Discovery to production lag Construction time Ramp up time Operating expenditure Employment Average grade Annual Production Resource

$AUD (M) Years Years Years $AUD per tonne FTE % FE MT MT

500 5 1.5 1.5 60 200 49.10 10 312

5000 7 3 2 55 450 46.40 30 951

10,000 10 5 3 46 900 60.40 60 2459

Although our objective is to estimate the return to the State government from the establishment of new mines, the decision to establish a mine is made by the private sector. We therefore model the decision regarding whether or not a discovery progresses to a mine as a private sector decision. For a discovery to progress to a mine the expected NPV of the project, from the private sector perspective, must be positive. For this decision we assume: a real discount rate of 12%; royalty rates for gold and nickel of 2.5%, and

7.5% for iron ore; payroll tax of 5.5%; and corporate tax of 30%. If, using these assumptions, progressing the discovery to a mine has a negative NPV, we assume the discovery does not become a mine. If a mine does proceed, we then value the return to the State government from royalties and payroll tax using real discount rates of 5%, 7%, and 9%. The State government raises the funds it spends to support exploration through taxation. The marginal excess burden for the GST is modest, while for stamp duty it is relatively high (Cao et al., 2015). For royalty taxes estimates range from a marginal excess burden of 70¢ (KPMG Econtech, 2010) to a marginal excess benefit of around 80¢ (Ergas and Pincus, 2014). For this exercise we assume that for every $1M spent on exploration investment $1.3M in taxation revenue must be raised. 3.2. Modelling results Modelling results are summarised in Fig. 2, and the individual plots demonstrate that, given the assumptions made, the expected return to the State government from investment to support greenfield exploration is likely to be positive. The total return to government flows from two sources: royalty taxes and payroll tax. The relative importance of these two sources of government revenue is shown in Fig. 3. As can be seen, the majority of the return to government comes from royalty payments. This is important, as due to the international nature of mining company ownership: “up to 75 per cent of government revenue from a royalty… …is a net transfer from overseas buyers and nonresident shareholders to Australian citizens” (Freebairn, 2015, p. 14). The return to government is, therefore, largely provided by those external to the local economy. The final element in understanding the benefit–cost ratio information shown in Fig. 2 is to understand the relative importance of the commodity price assumption and the private sector exploration multiplier assumption. To help understand the relative importance of these two factors Table 5 provides benefit–cost ratio information for a set of stylised scenarios. In Table 5 the row headings represents the price assumption, where low, mid, and high represent commodity prices at the 0.25, 0.50, and 0.75 point in the distribution; the column headings denote the private sector multiplier; and the values in the table represent benefit–cost ratios evaluated using a discount rate of 5%.

Fig. 2. Summary benefit–cost analysis.

J.J. Fogarty, S. Sagerer / Resources Policy 47 (2016) 78–86

85

Fig. 3. Relative importance of payroll tax and royalties.

By first reading down each column of Table 5 it can be seen that, for all multiplier values, moving from a low price assumption to a mid price assumption results in the benefit–cost ratio increasing by a factor of around 10; and moving from a mid price assumption to a high price assumption results in a further doubling of the benefit–cost ratio. Next, by reading across the columns it can be seen that moving from a low multiplier value to a multiplier value in the middle of the assumed distribution results in the benefit–cost ratio increasing by a factor of around 3–4; and moving from a low multiplier value to a high multiplier value results in the benefit–cost ratio increasing by a factor of around 6. Finally, it can be noted that regardless of the multiplier value, for the mid and high price assumption the benefit–cost ratio is always substantially greater than 1. Combined, the results suggest that the benefit–cost ratio is more sensitive to the price assumption than the private sector multiplier assumption. The table also makes it clear that it is only when all commodity prices are low and the multiplier value is at the extreme lower bound are benefit–cost ratios less than one observed. (Table 6) 3.3. Comparative perspective In terms of estimating a benefit–cost ratio for a similar government program, only one directly relevant paper was identified. Specifically, Scott et al. (2002) evaluate the return to Queensland Geological Survey actions to improve data quality; and across a range of scenarios and discount rates the authors report the upper and lower bounds for the benefit–cost ratio as 3.8 and 7.4. As ABS (2014b) indicates that the mining sector in Western Australia is more than twice as large as in Queensland;

Table 6 Relative importance of the price and multiplier assumption. Price/Multiplier

5

10

15

20

25

30

Low Mid High

0.4 3.5 7.1

0.7 6.9 14.1

1.1 10.4 21.3

1.4 13.9 28.4

1.7 17.3 35.5

2.1 20.8 42.5

and as the geology and institutions in Western Australian are very conducive to mining (Fraser Institute, 2014); it is reasonable to expect that the benefit–cost ratio in Western Australia would be higher than in Queensland. To put the result in context we can also look to a broader literature. For example, if we characterise government investment to support exploration activity as similar to general government research and development subsidies, then the broader return to research and development literature is relevant. Reviews of the return to research and development literature, for example Salter and Martin (2001) and Hurley et al. (2014), suggest that the government return to research and development is high (internal rate of return of around 10%). The general value of geoscience information literature also suggests the return to geoscience spending in general is positive (Haggquist and Soderholm 2015). So, although there is limited research that is directly comparable to our study, our findings are consistent with the broader literature surrounding the return to geoscience and the return to research and development.

86

J.J. Fogarty, S. Sagerer / Resources Policy 47 (2016) 78–86

4. Conclusion and qualifications Direct provision of basic geoscience to support exploration has long been government policy in Australia. Our findings show that for Western Australia, the return to the State government from this activity is positive. Specifically, as the real discount rate is varied from 5% to 9%, we find expected benefit–cost ratios of between 9.0 and 5.2 for State government investment to support exploration. A more recent development has been government provision of subsidies for exploration drilling. In terms of stimulating new private sector exploration activity, we find no evidence that drilling subsidies are less effective than traditional geoscience spending. We think this is due to the government drilling subsidy program being seen as a third party certification system for exploration companies. As such, the program works to mitigate the dishonesty externality in the market for exploration equity capital. We do however have a relatively short time series, and this limitation should be noted.

Acknowledgements The authors would like to thank Cameron Fraser and Margaret Ellis of the Department of Mines, Government of Western Australia, for assistance with data.

References Akerlof, G.A., 1970. The market for “lemons”: quality uncertainty and the market mechanism. Q. J. Econ. 3 (84), 488–500. Australian Bureau of Statistics (ABS), 2014a. Catalogue 8412.0: Mineral and Petroleum Exploration: Table 2 mineral exploration, (other than for petroleum) – Expenditure and metres drilled, Australian Government. Australian Bureau of Statistics (ABS), 2014b. Catalogue 5220.0: Australian National Accounts: State Account 2013–14 Table 4 and Table 6 Expenditure, Income and Industry Components of Gross State Product, Chain volume measure and current prices, Australian Government. Australian Bureau of Statistics (ABS), 2015. Catalogue 6302.0: Average Weekly Earnings Australian Government. Cairns, R.D., 1990. The economics of exploration for non-renewable resources. J. Econ. Surv. 4 (4), 361–395. Chamber of Commerce and Industry, 2013. REPS Major WA Projects List: July 2002 to December 2018. Cao, L., Hosking, A., Kouparitsas, M., Mullaly, D., Rimmer, X., et al., 2015. Understanding the economy-wide efficiency and incidence of major Australian taxes.

Working Paper No. 2015-01. The Treasury, Australian Government. David, A.P., Wright, G., 1997. Increasing returns and the genesis of American resource abundance. Ind. Corp. Chang. 6 (2), 203–245. Department of Finance, 1994. Flow Through Shares: An evaluation report, Executive Summary, Canadian Department of Finance. Dodds, D., Bishop, R.C., 1983. On the role of information in mineral exploration. Land Econ. 59 (4), 411–419. Duke, J.M., 2010. Government Geoscience to Support Mineral Exploration: Public Policy Rationale and Impact, prepared for the Prospectors and Developers Association of Canada. Ergas, H., Pincus, J., 2014. Have mining royalties been beneficial to Australia? Econ. Pap.: J. Appl. Econ. Policy 33 (1), 13–28. Freebairn, J., 2015. Reconsidering royalty and resource rent taxes for Australian mining. Aust. J. Agric. Resour. Econ. 59, 586–601. http://dx.doi.org/10.1111/ 1467-8489.12113. Fraser Institute, 2014. Survey of Mining Companies 2013. Geoscience Australia, 2014. Australian Mineral Exploration Review 2013, Australian Government. Government of Western Australian, 2014a. Western Australia Economic Profile, October 2014. Government of Western Australian, 2014b. 2014–15 Budget: Economic and fiscal outlook, Budget Paper. no. 3. Haggquist, E., Soderholm, P., 2015. The economic value of geological information: synthesis and directions for future research. Resour. Policy 43 (1), 91–100. Hogan, L., 2003. Public Geological Surveys in Australia, ABARE eReport 03.15 for the Department of Industry, Tourism and Resources, Canberra, August. Hurley, T.M., Rao, X., Pardey, P.G., 2014. Re-examining the reported rates of return to food and agricultural research and development. Am. J. Agric. Econ. 96 (5), 1492–1504. Khindanova, I., 2012. Impacts of Tax incentive programs on mineral exploration expenditures in Canada: an empirical analysis. J. Manag. Policy Pract. 13 (5), 81–87. KPMG Econtech, 2010, CGE Analysis of the Current Australian Tax System. Final Report 〈taxreview.treasury.gov.au/content/html/commissioned_work/downloads/KPMG_Econtech_Efficiency%20of% 20Taxes_Final_Report.pdf〉. Krautkraemer, J.A., 1998. Nonrenewable resource scarcity. J. Econ. Lit. 36 (4), 2065–2107. Peterson, F.M., 1975. Two externalities in petroleum exploration. In: Brannon, G. (Ed.), Studies in Energy Policy. Ballinger Publishing, Cambridge, pp. 101–114. Pesaran, M.H., Shin, Y., 1999. An autoregressive distributed-lag modelling approach to cointegration analysis Chapter 11. Strom, S., Chesher, A., Jackson, M. (Eds.), Econometrics and Economic Theory in the 20th Century: The Ragnar Frisch Centennial Symposium. Cambridge University Press, London. Productivity Commission, 2013. Mineral and Energy Resource Exploration, Productivity Commission Inquiry Report, no. 65, September 2013. Salter, A.J., Martin, B.R., 2001. The economic benefits of publicly funded basic research: a critical review. Res. Policy 30 (3), 509–532. Schodde, R., 2014. Challenges and Opportunities for Under-cover Exploration in Australia, Presentation to UNCOVER summit, 31st March 2014, Adelaide. Scott, M., Dimitrakopoulos, R., Brown, R.P., 2002. Valuing regional geoscientific data acquisition programmes: addressing issues of quantification, uncertainty and risk. Nat. Resour. Forum 26 (1), 55–68. SNL Metals & Mining, 2014a. Drill Results Database. SNL Metals & Mining, 2014b. Projects Database. SNL Metals & Mining, 2014c. Cash Cost Plus Database.