How accurate do you think the depreciation figures are for ChipMOS? Too high, too low, just right?
If I recall correctly, ChipMOS was depreciating equipment to zero in...6 years? When it was supposed to last a couple years beyond that. Also, much of that equipment is under-utilized right now so it's not getting worn out. However, technological equipment eventually becomes outdated regardless of whether it is used or not.
I'm trying to get an idea of how accurate the book value figure is since it's dropped so rapidly over the past year or two.
to the best of my knowledge you are asking the hardest non-BS question in stock analysis. can i recommend a more tractable alternative?
ultimately you don't need to know if stated book value is equal to liquidation value. what you need to know is, is it a meaningful normalizing constant for projecting future cash flows. that sounds like a harder question, but it is actually much easier.
first you want some industry aggregate data: EBITDA and tangible asset totals are what i use, but lots of alternatives exist. there's so much cyclical nature here that you HAVE to get some rough seasonal index values or you end up with noise.
OK, so now that you have some kind of empirical data, use it to deseasonalize EBITDA (or whatever) for the individual company you are studying. plot a time-series of EBITDA/ tangible assets. if there is no declining trend then you can reasonably conclude that depreciation allowance is roughly "fair". you can establish a valuation (as a multiple of book value) without ever quite establishing whether profits are really economic or accounting.
did i mention it sucks? yeah. but it is infinitely easier than trying to answer your question head-on. and all of that work tends to turn up a lot of interesting facts you wouldn't otherwise know to look for.
That does sound like it would render some interesting results, though I think getting rid of the noise would be difficult.
I prefer the equipment side. Let me give you an example:
I have an investment in a shipping stock. There are weekly reports for the price of ships, the number of new ships being produced and on order, the value for a ship if it is to be scrapped, the dayrates to hire a ship, etc. It's elementary to find liquidation value.
For a company such as ChipMOS that's a bit more complex. There's everything from real estate with negligible depreciation, to equipment that may last 5, 6, 10, or 12 years, to items that are useful for only the current generation of product. Luckily I'll be satisfied with just a rough estimate since there's such a drastic discrepancy between share price and book value.
As oort cloud suggested, I may contact investor relations. If they tell me anything useful I'll post it here.
While the depreciation schedule is of some interest, I thinks its even more important to think about how Chipmos is going to upgrade their equipment to test the new chips. Almost everyone in the industry has been investing the minimum in new equipment to conserve cash. One big problem is that those few companies that have upgraded are going to have a big advantage over those who haven't. Chipmos might survive two years or so, but if they don't invest in new equipment they won't make it much beyond the time their old equipment wears out.
I am not an expert on how long dram chips last, but I think that the chipmos' six year scheudule should serve as a good guideline to when they think its useless.
My understanding is that IMOS is using five year straight line depreciation or about 20% a year. Dram testing equipment becomes obsolete more easily, especially those machines for DDR2 as 1333 MHz DDR3 will become main stream next year. Their DRAM testing machine may be good for low end DDR3 like 1000 MHz DDR3. Wafer sort and packaging equipment will last longer than their depreciation life. As Spansion is coming out of Chapter 11 in Q4, the capacity utilization of wafer sort machines may improve. IMOS is trying to get business form Numonyx. Recent news is that Numonyx will start making chips in house instead of outsouring that to Elpida. The problem is that IMOS has become a second tier memory backend company. It can only get overage business that Powertech Technology Inc. utilizes its full capacity. When time is good, the difference is not too big. When time is bad, Powertech is making money while IMOS loses its shirt.
My understanding is that companies use the shorter of the time the equipment physically wears out or the time that the equipment becomes functionally obsolescent. Six years seems about right for the entire life cycle of a dram, so I'd say that that six year depreciation sounds fine.
However, when equipment is underutilized in a recession then companies may have to do what's called an impairment test where they look at the drop in value over time. If the impairment is significant, then they have to reduce the value of the equipment. We only get to observe impairments on financial statements if they're pretty large. My suspicion is that the drop in value for the equipment is due to them writing off the value of some of the equipment early. Its probably impaired, but not enough to be separately reported.
A suggestion is that you contact Chipmos directly and ask your question. While they seem to be a bit clueless about publicizing their positions, they are surprisingly honest about what's going on. Its one of the reasons I've held onto the stock for the last two year.