Pan African Resources — A second glance at the first half

PAF’s interim results to 31 December 2017 revealed group-wide gold production of 85,282oz at a cash cost of US$1,099/oz. Whereas this represented a decline in gold output of 6.9% compared to the prior year period (H117), in fact it represented a 4.4% increase compared with the period immediately preceding it – ie H217. The H118/H217 increase could be attributed to higher aggregate throughput and recoveries. Combined with higher prices in both rand and dollar terms, this resulted in a 26.7% increase in on-mine revenue, albeit substantially offset by a similar increase in aggregate costs of production and a disproportionately high increase in the depreciation charge. Nevertheless, mining profit was 38.3% higher than in H217. With an absence of any material exceptional items, such as hedging mark-to-market, profits/(losses) on disposals, impairments and contributions from associates, this translated into a 432.7% positive variance in profit before tax (US$4.3m profit vs US$1.3m loss) and EPS were then only lower on account of an effectively exceptional tax credit in H217 (see Exhibit 1, below):

Excluding the expansionary Elikhulu project, group aggregate capex declined by 18.9%, or ZAR41.6m, across the company’s four producing assets on account of the completion of the Barberton Tailings Retreatment Project (BTRP) cyanide detoxification plant and Fairview’s ventilation refrigeration and infrastructure. Otherwise, the six-month period was characterised by delays in developing into Fairview’s high-grade 272 and 358 platforms, 11 days (equivalent to 3,000oz) in lost production owing to industrial action by employees and protests directed by community pressure groups and a portion of the high-grade 101 platform at Fairview’s MRC orebody being sterilised by an unanticipated geological roll (all at Barberton). At the same time, performance at the BTRP continued to be affected by the metallurgical consequences of the transition from the Bramber to the Harper dumps (first apparent in H217 – see Exhibit 5).

In rand terms, PAF’s gold cost of production (excluding realisation costs) increased by 5.4% to ZAR1,228.0m relative to H117 (cf up 14.4%, to ZAR1,165.6m in H117 vs H116). Superimposed onto a 0.3% decline in total tonnes processed, this translated into a 5.4% increase in group-wide unit working costs, to ZAR678/t. Superimposed onto a 6.9% year-on-year decline in gold produced, it translated into a 13.0% increase in unit costs of production, to ZAR473,187/kg (or an 18.0% increase in US dollar terms, to US$1,099/oz after forex considerations). Including realisation costs, PAF’s gold cost of production increased by 11.2% to ZAR1,225.2m relative to H117 (cf up 16.4%, to ZAR1,193.3m in H117 vs H116). Notable features were:

All of the group-wide increase in gold production in H118 could be attributed to the recovery in output at Evander after a 55-day underground mining hiatus in H217, during which steelwork at the 7A, 7 and 8 shafts was refurbished. As a result, adjusted EBITDA from PAF’s underground operations (ZAR107.6m) was once again greater than that from its tailings retreatment projects (ZAR106.4m) – although it remained the case that the ETRP and the BTRP were PAF’s second and third most profitable business units, overall, behind Barberton (underground), but ahead of Evander (underground).

Evander’s recovery in H118 vs H217 was exacerbated by the fact that the latter period contained a 55-day mining hiatus, during which steelwork at the 7A, 7 and 8 shafts was refurbished. Even so, comparing its performance in H118 vs H117, tonnes milled and head grade were both higher, while total cash costs, expressed in ZAR per tonne milled, were lower (see Exhibit 2).

During the six-month period under review, the existing 8 Shaft pump column (the only one that was operational in the period) experienced a number of water bursts, which contributed to lost production, but will cease to be a problem once EGM’s refurbishment programme is completed in April. In the meantime, development of the new high-grade D raise is being accelerated with the intention of making it available for production this month to contribute to increased mining flexibility and access to higher-grade areas of the 8 Shaft orebody. A table of Evander’s performance by half year period is provided below in addition to our forecasts (column eight) and a number of alternative contingencies (last three columns):

While EGM recorded a profit in H118 for the first time since H116, it is notable that, given the recent strength of the rand, our headline financial forecasts for Evander remain under pressure in H218 (column eight of the above table). Four possible remedies exist to correct this; two are partially within management’s control and two outside it:

For Evander to cover its all-in costs in H218, including capex (in contrast to our ‘base case’, which is provided in column eight of Exhibit 2):

Note that, of the above, we regard two of these contingencies (grade and costs) as somewhat within management’s control and two (the gold price and the rand) as without it. Of the four, we maintain that the most readily achievable is our unit working cost target, followed by rand depreciation, grade and the gold price. In addition to cost cuts already made, on 27 February, PAF confirmed that Evander is in a consultation process with its labour in terms of section 189 of the South African Labour Relations Act, 66 of 1995. Section 189 (1) of the Labour Relations Act provides that, before retrenching, employers must consult any person whom they are required to in terms of the collective agreement.

While we believe it is entirely conceivable that EGM can return to a profitable and cash-flow positive position, it is possibile that, in the absence of any tangible change by the end of its financial year, management may decide to place 8 Shaft on either care and maintenance or to close it entirely. The valuation consequences of such a decision are considered more fully in the ‘Sensitivities’ section on page 17 of this report. In the meantime, however, while it is notable that EGM will account for c 36.7% of PAF’s group-wide gold output in FY18 on current forecasts, it will account for only c 0.6% of gross profits. Moreover, even if costs return to our forecast, steady-state level (as depicted in Exhibit 2), in the long term, we estimate that EGM’s EBITDA margin will average only 14.7% over the life of operations and will not exceed 20.8%; ie it will be the lowest margin of PAF’s operations. In the event that operations do cease at Evander at the end of the current financial year therefore, the effect on our overall valuation of PAF (see Exhibit 13) is only 2.55p, or 13.0%, excluding exceptional costs.

In recognition of the possibility that operations at Evander’s existing infrastructure could cease in the foreseeable future, on 28 March, PAF announced it will reassess the feasibility study on the Egoli project (see page 10) as a standalone project. This is expected to be completed by the end of the current financial year.

Underground production at Barberton declined by 2,312oz compared with H117 owing to a lack of grade flexibility in the Fairview MRC orebody as well as disruptions from pressure groups, community unrest and a combination of protected and unprotected strike action, which resulted in 11 days of lost production (equivalent to c 3,000oz of gold output) as employees were prevented from reporting to work. Note that, the source of the frustration within the local community is reported to be unrelated to PAF, but instead a product of the general dissatisfaction with government service delivery, inter-union rivalry and unemployment. This also compares with eight days of lost production in H117 as a result of the Department of Mineral Resources serving six Section 54 regulatory notices at Barberton and six days lost to community unrest. In addition, a portion of the high-grade 101 platform was sterilised as a result of an unanticipated geological roll.

Exploration drilling has confirmed at least a 200m down-dip extension of the high-grade 11-block of the MRC orebody. Consequently, since H217, work has been underway to develop additional production platforms to expose additional high grade panels to increase mining grades and flexibility. While there were some delays in putting in the required development, this initiative was successfully concluded towards the end of H118/start of H218, with the establishment of the 272 and 358 platforms. The 272 Level platform was inspected by the author of this report on 1 February and was observed to be operational. In combination with the deeper 358 platform, this will allow management to ‘cycle’ high grade production from the two levels in future such that output is of a much more consistent grade than in the recent past. Subsequently, on 2 March, PAF announced that the average grade for the Barberton complex had risen from 8.7g/t in H118 to 11.5g/t in the February production month (a 32.2% increase) and that the residual, non-high grade mineralisation of the 11-block had a 10.5g/t grade. As a result, on 28 March, it further announced that it expected Barberton Mines (including the BTRP – see below) to “produce approximately 50,000oz of gold in the second half of the 2018 financial year”.

Development to the next high-grade platform (the 256 platform) will start early in FY19 ahead of its entering production early in FY20. Average grades at the platform are estimated by management to be 28.4g/t over 5.2m of width and 95m of strike length. As well as improving mining flexibility, this platform is expected to sustain gold production from the MRC section for approximately five years.

Production at the Barberton Tailings Retreatment Project (BTRP) in H118 was below the level of both H117 and H217 owing to the shift in operations to the Harper dump, after the exhaustion of the Bramber dump. As a result, head grades declined materially, to 1.40g/t (see Exhibit 5, below). In addition, the Harper dump material has a larger coarse fraction, which resulted in processing difficulties and a reduction in plant recoveries.

A regrind mill is on track to being commissioned in the last week of April to reduce the Harper dumps’ coarse fraction material, which should have the combined effect of both improving material handling and recoveries. Otherwise, nameplate capacity at the BTRP is 100ktpm and management is working towards achieving this level to sustainably produce 10,500oz in each semi-annual period. In addition, it has also recently been investigating the potential to reduce retention times, while maintaining metallurgical recoveries, in which case our longer-term forecasts (see pages 14-15) may ultimately prove to be conservative. More generally, approximately two-thirds of the tailings being treated at the BTRP originate from the Fairview concentrator at a grade of c 1.6g/t, while the remaining third have the same origin but are supplemented with material from the Biox plant at a grade of 3-10g/t. As a result, a substantial portion of the resources being treated by BTRP exist at a grade in excess of 1.6g/t and there is therefore ample opportunity for management to pursue grade-optimised strategies in future to improve financial returns to shareholders.

Effectively, the ETRP represents a substantial pilot plant, designed to prove recovery and cost parameters, before the development of the much larger Elikhulu project (see below).

Production at the ETRP reduced in H118 compared with both H117 and H217. However, this could be attributed to higher levels of surface material being processed by the plant during those periods while 7 Shaft infrastructure was being repaired. Nevertheless, the plant achieved a high level of capacity utilisation (91.0% vs a management target of 150-160ktpm, or 75-80%, on a sustainable, long-term basis) and a high level of metallurgical recovery (56% vs 45%), which contributed to costs continuing to be lower than their forecast long-term level of ZAR417,963/kg.

The grade of the dam being re-mined at the ETRP is 0.3g/t. However, the operation is commercially viable given its ability to fill unutilised capacity in the Kinross plant such that it therefore attracts only incremental operating costs (eg 14 additional employees). Re-mining is being conducted without breaching the dam wall and the tailings are being redeposited at Winkelhaak, thereby simplifying the environmental requirements with respect to re-filling the Kinross dam at a later date.

In the meantime, however, management will continue to source toll-treatment material with higher grades than the ETRP’s reserve and resource grades, thereby taking commercial advantage of the fact that it is the only retreatment operator in the area and is therefore (effectively) the buyer of choice – or even the buyer of last resort – for tailings assets destined for retreatment in the region.

Construction at Elikhulu is reported to be progressing ahead of schedule and below budget, with first gold now expected from the complex in August 2018 (vs Edison expectation H219 – ie the first half of CY19) and the ramp-up to full production anticipated to take no longer than two months. Compared with the ETRP’s 2.4Mtpa processing plant capacity, Elikhulu will process 12Mtpa to produce c 50koz per year, at an initial capital cost of c ZAR1.7bn (of which ZAR671.4m has been expended to date) and ongoing cash costs of c US$398-504/oz and all-in sustaining costs below US$650/oz. The re-mining contract for the project has been awarded to Fraser Alexander (which already has a similar contract at the ETRP) and incentivises it to process more than 1Mtpm.

In addition, a feasibility study has now been completed by DRA, which confirms the viability of a post-commissioning capacity increase at Elikhulu to allow the plant to process the c 200kt of tailings per month currently assigned to the ETRP for “limited” capex and with resultant cost and throughput benefits as the same feed is exposed to Elikhulu’s lower cost structure and higher metallurgical recoveries..

In addition to Elikhulu, PAF has three other significant growth projects, namely Egoli (formerly the Evander Mines 7 Shaft No. 3 Decline and 2010 Pay Channel project), the Royal Sheba project and the Barberton Mines Sub-Vertical Shaft project at Fairview.

The 2010 Pay Channel contains an estimated 2.95Moz of resources (of which 1.03Moz are in the measured and indicated categories) at an average grade of 9.75g/t and is c 4.5km in tramming distance from 7 Shaft, which is used by EGM for hoisting to the Kinross metallurgical plant (cf 8 Shaft, which is c 12km distant). The previous operator, Harmony, had previously developed the 7 Shaft mine working towards the 2010 Pay Channel, but discontinued the initiative in 2009, allowing the controlled flooding of the development ends and 7 Shaft’s No. 3 Decline, from 21 Level to 18 Level. To date, two boreholes have successfully been drilled into the 2010 Pay Channel, intersecting the Kimberley reef at a depth of c 2km. The first yielded a reef intersection with a width of 49cm and a grade of 36.04g/t (a metal content factor of 1,766cm.g/t), while the more recent recorded a width of 6cm and a grade of 36.8g/t (a metal content factor of 221cm.g/t). Additional drilling deflections are being performed to further delineate the orebody.

In the meantime, a feasibility study on the project has been completed with the following major features:

We are not privy to the detailed financial model for Egoli. From the information provided, however, it is able to make a preliminary valuation of the project on a post-tax basis of ZAR929.7m at our updated long-term gold price assumptions (see Exhibit 8) and prevailing forex rates, which equates to 53.4% of PAF’s pre-tax NPV10 of ZAR1.74bn, above (among other things, reflecting a relatively high average gross margin of 47.7% for tax purposes), or ZAR0.52 (4.4 US cents, or 3.2p) per share. Note that the figures provided above also imply a unit cash cost of ZAR2,163 per tonne milled, which approximates our targeted costs for EGM’s existing operations on the Kinross Pay-Channel (see Exhibit 2 and pages 4-5).

While the study inevitably conceives of the Egoli project as an integrated constituent of the broader Evander complex, in view of the section 189 process underway there (see page 6), on 28 March PAF announced it will reassess the feasibility study as a standalone project and that this process is expected to be completed by the end of the current financial year.

The Fairview mining operation is restricted by the hoisting capacity of its No. 3 Decline, which is used to access workings below 42 Level and the high grade MRC 11-block orebody. Currently, the decline is used to transport employees and material as well as for rock hoisting and, with no modifications, future mining at depth will be compromised by increased travelling distances, reduced employee face time and a lack of sufficient capacity to ensure both adequate ore replacement and exploration development. With this in mind, PAF has now completed a study with DRA to investigate the feasibility of constructing a raise-bored, sub-vertical shaft from Fairview’s 42 Level to 64 Level and, potentially, in future, to 68 Level (resources extend down at least to 74 Level). The sub-vertical shaft will then be used to transport employees and material to the working areas, while No. 3 Decline will be used exclusively for rock hoisting, thereby significantly increasing overall capacity and production from this high grade mining area as well as improving ore handling efficiencies.

Estimated capex for the project (including contingencies) is ZAR105m (US$8.9m, or £6.4m at prevailing forex rates) and would result in estimated, additional output of 7,000oz gold per year, which “can be optimised further to more than 10,000oz per year.”

Assuming that construction takes place in CY18 and CY19 and that production begins in CY20 with cash costs of c US$800/oz over 15 years at the gold prices indicated in Exhibit 8 and prevailing forex rates, we estimate that this project could be worth in the order of US$10.5m (0.58c, or 0.42p, per share) to PAF, rising to US$16.4m (0.91c/share) in the event of optimisation (at our standard 10% discount rate).

The Royal Sheba orebody forms part of the Barberton mine and, before 1996, was mined on a small scale to a depth of 340m below surface. In FY10, a concept study was completed with the aim of re-opening the mine as a larger, standalone operation. The study concluded that the project was viable, but required significant capex in the form of a new shaft system from surface and the construction of a new gold plant. Since then, however, three significant synergies have been identified that suggest that capex can be materially reduced. These include:

Management believes that Royal Sheba has the potential to produce c 30,000oz of gold per annum, to which end it has commenced a drilling programme of 14 holes from surface, totalling 12,000m (an average of 857m/hole), to increase geological confidence in the orebody and to contribute towards a feasibility study, which is expected to be completed in CY18. In the meantime, management has developed a full three-dimensional model of the Royal Sheba ore body, incorporating the structural, lithological and mineralisation components of the deposit, which has resulted in an increase in resources both adjacent to and below the current mine infrastructure and a formal upgrade in its mineral resource estimate, as follows:

The updated mineral resource estimate considers 10 domains within the deposit and is staged over a larger down-dip extent than previously, owing to the variogram model parameters applied. Relative to PAF’s prevailing resource multiple of US$7.14/oz, a 358.8koz resource increase has a pro-rata value of US$2.6m.

We have has updated our gold price forecasts since its last note on PAF. These are set out in detail in our report, Mining overview: Unlocking the price to NPV discount, published in November 2017, and are summarised below in real terms:

Note that, Edison’s average (real) gold price forecast over the course of PAF’s mines’ lives, from FY19-39 inclusive, is US$1,384/oz.

After producing 85koz of gold in H118, our updated group-wide production estimate (as per Exhibits 2 and 4-6) in FY18 is 178koz, which correlates closely to management’s guidance of 177-181koz (thereby implying production of 92-96koz in H218):

Within this context, our detailed financial forecasts for PAF for H218 and FY18 are therefore now as follows.