Pan African Resources — Update 9 December 2015

Considered on a standalone basis, Pan African’s FY15 results were characterised by a large increase in tonnes milled and processed, primarily as a result of a maiden contribution from the ETRP, thereby firmly establishing it as a retreatment operator as well as an underground miner. In local currency terms, the gold price was higher in H215 cf H1, owing to the depreciation of the rand, although this was offset by a decline in gold sold, such that FY15 revenue was effectively flat
(-3.8%). Coupled with continuing cost increases on a gross basis, this resulted in a degree of margin compression and a consequent 25.5% year-on-year decline in adjusted EBTIDA.

For H215 specifically, precious metal sales were £7.4m less than our prior expectations, owing in part to an under-sale of material relative to production and the continuation of the low-grade mining cycle at Evander for longer than expected. However, this was partially offset by production costs also being £4.3m less than expected. Royalty costs and taxes were £1.1m and £1.6m better than our expectations, partially offset by a £1.5m negative variation in net finance costs. Overall, H215 net profits were £2.2m less than our expectations in H2. Nevertheless, this still represented an increase cf H1, as shown below.

In general, Pan African’s surface retreatment operations outperformed expectations, driven by higher than expected metallurgical recoveries. Underground operational performance was inevitably more affected by the inconsistent electrical power supply in South Africa and also the Section 54 stoppage notices imposed on both Barberton and Evander by the DMR during the year. Nevertheless, cash costs remained below US$1,000/oz and both reportable (RIFR) and lost time (LTIFR) injury frequency rates remained on a downward trajectory.

Pan African’s immediate focus is now to increase output to 225,000oz per year (vs 176koz in FY15) and to improve its cash cost profiles.

At Evander, the six months from December 2014 to June 2015 was characterised by a slower than expected recovery from the lower-grade mining sequence. As a result, the (implied) underground head grade was 4.92g/t compared with a previous forecast of 5.25g/t. This coincided with a decrease in tonnes milled from underground sources, owing to challenges related to underground mining operations and infrastructure constraints (especially conveyor belt availability), allied with power interruptions and DMR Section 54 stoppage notices, which resulted in three days of lost production in H1 and six days in H2. As a result, there was an increase in unit costs, from ZAR1,230/t in H115 to ZAR1,795/t in H215. However, this largely reflected a sharp, 65.9% decrease in the number of tonnes treated from surface sources. Excluding this effect, the likely unit cost of mining from underground increased only 5.6%, from ZAR2,250/t to ZAR2,376/t (cf our target of ZAR2,159/t). As a result, gross total cash costs fell materially for the first time since H213, to ZAR451.9m.

Evander has implemented remedial action regarding conveyor belt availability, including the replacement of 4.4km (out of c 5km) of belting and the introduction of a planned maintenance function (cf previous breakdown maintenance), which has resulted in an improvement in conveyor belt availability from 60% to 80%. The mine has also improved the monitoring and pump infrastructure of its 8 shaft dewatering pumps, including the introduction of a dam cleaning schedule, to reduce the risk of shaft flooding.

In rand terms, the major contributors to cost pressures on a year-on-year basis were:

Evander is supplied by only one feeder cable from Eskom (vs four at BGMO). This increases the operational risk of being affected by power outages. However, Evander has also been designated as a ‘key industrial customer’, with the result that it is the last affected operation in its region and has never experienced an uncontrolled power cut. Unlike Barberton, however, it lacks the capacity to support itself on self-generated power, with the exception of the winders, emergency pumps and fans.

Level 25 at 8 shaft has now been established and raise lines installed, providing access to additional high-grade panels and creating flexibility in the mining operation, which will allow management to manage and blend the ore mined towards achieving a target of 110,000oz of annual gold mined at EGM. While operations are ramping-up at 25 Level, the 24 Level return airway has been converted to create a tramming loop, with the result that the majority of crews are now working in the high-grade area of the mine – this is expected to remain the case for at least the next two years.

As a result of the migration of productive activities from the 23 to 25 Level, management is forecasting an increase in both tonnes hoisted and head grade at 8 shaft to 5.68g/t in H116 and 6.75g/t in H216.

Face-grade and head-grade forecasts are supplied by management, from which we have calculated the average head grade for the relevant half-year periods.

Both face grade and mined grade are part of a mine’s ore flow controls, controlled by Evander’s detailed Short Interval Control (SIC) system. Face grade, in particular, is the average grade of the tonnes mined on the reef horizon at stoping widths and includes all additional waste, eg winch cubbies, gulley and footwall waste. The mining grade is the face grade adjusted downwards with all the other waste produced by mining. In an ideal situation, a mine call factor (MCF – defined as the ratio of the total quantity of recovered and unrecovered mineral product after processing compared to the amount estimated to have been in the ore based on sampling), can be applied to the average mining grade (if all sources of ore are considered), to equate to a head grade. Thus, if the process of sampling, assaying and tonnage measurements in a mine and plant were to be perfect and there was no mineral loss at any stage during handling and processing, the MCF would be 100% and the mined grade would equal the head grade. Any attempt to infer a head grade from a mined grade at Evander 8 Shaft also therefore needs to consider waste, on-reef mining and the MCF – only one of which (the MCF) is readily available. In this case, the historic MCF at Evander is 63%. Recently, it is reported to have been over 70% and as high as 89%. We also regard it as significant that it has been on a broadly rising trend.

More simply, it can be observed that, over the period covered by the graph in Exhibit 3, the head grade has typically been between 55.8% and 36.4% of the face grade (with an average at 43.8%), which would imply a potential range of head grades of 7.75-5.06g/t in H116 and a 8.33-5.44g/t in H216.

Future operations at Evander will involve advancing from 25 to 26 Level in approximately 12 months’ time, increasing the number of conveyor belts from 11 (over 5km) currently to 12, and proportionately increasing the amount of associated maintenance required. However, it will also improve access to more high-grade panels and allow management to selectively reduce mining at the lower-grade extremities of the Kinross pay channel.

In the longer term, management believes that a grade of 7.2g/t is sustainable. Note that, had an underground head grade of 7.2g/t prevailed in H215 (and applying the group’s H215 marginal tax rate to the incremental profits), we estimate that it would have added c 0.52p to both Pan African’s EPS and headline EPS in H215 (ie 1.04p on an annualised basis).

While the performance at BGMO improved in H215, it continued to exhibit the after-effects of the contamination of the Biox plant with oil at Fairview. Although metallurgical recoveries from the Biox plant recovered to pre-contamination levels, overall throughput was c 91% of capacity. At the same time, the head grade was affected by face-grade availability at both Sheba and Consort mines. However, cash costs were also well controlled with the result that gross profits are likely to have increased during the second six-month period in H215 in both ZAR and USD terms.

Principal sources of cost inflation over the year (in rand terms, year-on-year) were:

Note that processing costs fell 1.6% during the year. Capital expenditure similarly declined by 25.4% to ZAR112.6m – almost all of which is now classified as either ‘maintenance’ or ‘development’ after the conclusion of the BTRP’s construction.

Operational and maintenance systems have been implemented to mitigate the risk of future oil contamination of the Biox plant. At the same time:

Finally, exploration drilling confirmed the down-dip extension of the high-grade 11-block of the MRC orebody by a further 170m, thereby increasing the life of BGMO’s operations by 5.3%, from 19 to 20 years (company estimate).

Although it declined in H2 vs H1, the head grade at the BTRP was still higher than expected. Higher metallurgical recoveries (average 57% in FY16, but higher in H2, vs a planned 56%) also contributed to higher production, while simultaneously mitigating the decline in grade from a unit cost perspective. As a result, gross profits expanded in both ZAR and GBP terms in H2 (although they declined in USD terms, owing to the effects forex rates), while unit costs of production remained below US$500/oz. Capital expenditure declined to zero, as expected.

Throughput is anticipated to increase towards design capacity throughout FY16. For the moment, we are anticipating a continued moderation in grade and metallurgical recoveries. However, management has recently been investigating the potential to reduce retention times, while maintaining metallurgical recoveries at relatively high levels. To the extent that it is successful, our forecasts may prove to be relatively conservative.

In the longer term, approximately two-thirds of the tailings being treated at the BTRP originated from the Fairview concentrator at a grade of c 1.6g/t. The remaining third are from the same origin, but are also 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 exists at a grade in excess of 1.6g/t and there is ample opportunity for management to pursue high grades to increase financial returns.

Notable in Pan African’s H215 results was a maiden contribution from the ETRP, which achieved commercial production in early H215 compared with expectations of H116, and therefore contributed around four months of output and revenues to group results.

Pro-rata, the ETRP achieved 89.9% throughput capacity utilisation. In addition, metallurgical recoveries were higher than expected at 54% cf 42% expected. In part, this could be attributed to the processing of additional sources of higher-grade surface feedstocks. Even excluding this, however, recoveries were reported to be 48%, with the result that unit costs of production for the period were below their ZAR280,000/kg target.

The grade of the dam being re-mined at the ETRP is 0.3g/t, which is low in absolute terms. However, the operation is commercially viable given its ability to fill unutilised capacity in the Kinross plant and the fact that it therefore attracts only incremental operating costs (largely for 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 (ie the Environmental Impact Assessment) with respect to re-filling the Kinross dam at a later date.

Effectively, the ETRP represents a substantial (10,000oz per year) pilot plant, designed to prove recovery and cost parameters, before a decision is taken on a much larger operation. Named Elikhulu, this would process 12Mtpa (cf the ETRP’s 2.16Mtpa), but would require material independent capital expenditure and would need to support an independent cost base (cf the ETRP, which effectively supports only an incremental cost base).

Phoenix’s performance during the year was transformed by IFM’s resumption of mining at its underground Lesedi Mine, which provided sulphide material for treatment in the CTRP (vs oxide material from Skychrome previously). This, in combination with an improved reagent suite and the processing of higher sulphide content tailings contained in number 4 dam, resulted in a materially improved plant recovery and financial performance. As a result, Phoenix’s adjusted EBITDA for the year was reported to have been ZAR27.7m, which was in line with (and a slight improvement on) management’s prior internal profit forecast of ZAR24m.

Of note is the fact that, while gross costs and unit working costs rose, unit costs of production fell as a result of the operation recovering to its targeted grade profile.

Year-on-year contributors towards upward cost pressures in FY15 were:

On 27 August, International Ferro Metals (IFM) announced that, as a result of deteriorating business conditions, its South African subsidiary had entered a Business Rescue programme – a statutory means of enabling a financially distressed company to continue business under the supervision of a Business Rescue Practitioner, protected from its creditors. Phoenix Platinum is on IFM’s property and a portion of the feedstock for its operation (currently c 20%) originates from tailings arising from IFM’s processing activities. It also sources electricity, water and certain other services from IFM. At this stage, management states that it "is not in a position to fully assess the impact of the Business Rescue proceedings" on Phoenix Platinum. In the interim, Phoenix is investigating the feasibility of sourcing material from its Elandskraal and Kroondal tailings dams to maintain production and to mitigate any shortfalls arising from IFM (note that on 29 June Phoenix signed a new agreement to secure the PGE rights to the Elandskraal surface resource).

In June 2015, Pan African entered into agreements to acquire the Uitkomst colliery, near the town of Utrecht in KwaZulu-Natal, from Oakleaf Investments Holding 109 Proprietary and Shanduka Resources for a cash consideration of ZAR200m (£9.3m at current exchange rates).The colliery is already operational with 25.7Mt of resources (of which 86% is in the measured and indicated categories) selling c 0.4Mtpa of high-grade thermal, export-quality coal with metallurgical applications to both the domestic and the export markets.

The acquisition remains subject to DMR approval. However, it is anticipated to be immediately earnings- and cash-flow accretive. Until it closes, however, it remains excluded from our forecasts.

Historically, Pan African has had good relations with the unions. Evander negotiates as part of the collectivised bargaining process, under the auspices of the South African Chamber of Mines, while Barberton negotiates separately. During the 2013 wage round (which covered FY14 and FY15), Pan African was the first entity to reach a settlement.

On 13 October 2015, the company further announced that it had again concluded multi-year wage agreements with both the NUM and the UASA to the effect that:

As has been well documented, South Africa is currently suffering from a shortage of effective generating capacity as older power stations on the historic Eastern Transvaal coalfields approach obsolescence.

In response, Eskom has commissioned two new power stations in South Africa’s Limpopo and Mpumalanga provinces. Known as Medupi and Kusile, respectively, each plant will have a generating gross capacity of c 4,800MW, with Medupi representing the largest investment in Eskom’s 84-year history and the first base-load project to be built in the country in 20 years. The combined output of the plants represents c 23% of the country’s current power generating capacity.

However, the completion of Medupi is taking longer than anticipated. Whereas Eskom estimated that the total duration of the project would be no longer than four years at the time of its commissioning, over seven years later only one unit of 794MW capacity has been built (out of six) and the project remains some way from completion. While the news of first electricity and its synchronisation into the grid in March is positive in the context of South Africa’s electricity tribulations, it represents only one-sixth of design capacity and only 1.9% of Eskom’s current capacity of c 42,000MW. Moreover, capital expenditure on the two new power stations has exceeded the budget, such that Eskom has needed to turn to the South African Treasury for additional funding. As a result, in addition to the three 8% per year increases agreed between Eskom and the regulator, NERSA, in 2013, there was one further increase of 4.69% from 1 April 2015. Nevertheless, at c ZAR0.85-0.90/kWh (c US$0.060/kWh), electricity remains cheap in South Africa relative to the rest of the world – not least as a result of the depreciation of the rand, which has resulted in the equivalent US dollar price falling from US$0.066/kWh in the past six months. In addition, all operations at the company are engaged in a process of load shedding and power clipping aimed at minimising electricity consumption wherever possible. Note that the cost of self-generating power at Pan African’s mines is typically 3-4x the cost of grid power.

At least until the two power stations are commissioned and synchronised into the grid, South Africa appears likely to continue to experience intermittent load shedding and power outages. However, this is almost invariably directed towards residential, rather than industrial consumers. By contrast, industrial consumers are typically required to reduce their consumption by 10% or, at worst, 25%. As a consequence of this situation, it is supposed that the South African government will look more favourably on independent power producer-generating solutions for South Africa’s future electricity needs. It will also focus on the previously largely unexploited resources of the Waterberg coalfield in the Limpopo province for its energy source (note, readers are directed to Sibanye’s recent private sector initiative in this region).

We have updated our forecasts for FY16 based on the estimates made for each of the precious metals’ operations and highlighted individually (note that forecasts continue to exclude the Uitkomst colliery, as stated above). On the basis of these assumptions, our detailed financial forecasts for PAF for H116, H216 and FY16 are as follows.

Note that our FY16 headline EPS forecast of 1.43p per share compares with a mean consensus estimate of 1.56p, within the range 0.36-2.00p (source: Bloomberg, 10 December 2015).

Additional key changes to our long-term forecasts compared with our last note (and in line with management guidance) are as follows:

Updating our long-term forecasts to reflect these changes, our absolute value of PAF increases from 22p to 25p (in part as a result of advancing one year from 1 July 2014 to 1 July 2015), based on the present value of our estimated maximum potential stream of dividends payable to shareholders over the life of its mine operations (applying a 10% discount rate).

The above profile assumes that the grade profile at Evander will increase to in excess of 7.00g/t from FY17. We assume gold prices of US$1,372/oz in FY17 and US$1,378/oz in FY18.

Three key sensitivities affect our short-term earnings forecasts, namely the Evander grade, Evander costs and the gold price:

In the longer term, the key sensitivity affecting our forecasts is the real gold price. Currently, our forecasts for FY17-39 are US$1,454/oz and for CY17-39 are US$1,466/oz.

If the gold price remains at US$1,070/oz in real terms, however, our absolute value of PAF halves from 25p to 12.5p (all other things being equal), as shown in the chart below:

Note that average long-term EPS of 1.67p/share for FY18-28 is approximately consistent with FY15 EPS of 0.64p plus 1.04p (0.52p in H215 annualised) if Evander recorded an underground head grade of 7.2g/t.

Of as much significance, PAF remains notably cheap (based on our updated estimates) when compared with its South African peers in Exhibit 11, below.

Therefore, on both consensus and our forecasts, Pan African is consistently cheaper than its peers in terms of yield (100% of instances considered over the two years), P/E ratio (75% of instances considered) and EV/EBITDA (at least 62.5% of instances considered).

Pan African had £18.0m in net debt at 30 June 2015, which is an improvement on our prior forecast of £24.0m. Debt is financed via a ZAR800m revolving credit facility (£37m at current exchange rates), which can be expanded to ZAR1,100m (£51.0m). Net debt currently equates to a gearing (net debt/equity) ratio of 12.3% and a leverage (net debt/[net debt+equity]) ratio of 10.9%, but is forecast to improve to £3.9m under the effect of the improved cash flows engendered by the improved operational performance described above before consideration relating to the acquisition of Uitkomst (assuming DMR approval) of ZAR200m (£9.2m at current exchange rates).