Valuation identifies when a company is trading for less than fair value. We use our fundamental knowledge to build a forecast, we use that forecast to value the company, and we win if this forecast is more accurate than the market.
Valuation isn’t a science. Bankers and consultants write books about valuation as a science because their job relies on selling precision, as opposed to actually attaining it. Keynes’ analogy of the stock market as a beauty contest makes clear that even if we knew the future with certainty, we still have to predict how others will feel about the future.
This post is an attempt to provide a starting point to this complex exercise. The simplicity is not to mislead but to give you the tools needed to cut your own path through the jungle.
I don’t like hard rules but everyone needs to start somewhere:
Buy stocks when the 5 year forward P/E is less than 10. If today’s share price is £1, the company will have to earn 10p five years in the future.
Where did I come up with this number? Nairn and Davis’ Templeton’s Way With Money.
Understanding whether your valuation is actually meaningful relative to the market price is hard. To make money, the stock has to beat expectations in some way but what are expectations? Analyst estimates are usually not quite right. It is easy to make mistakes.
This technique provides a starting point, focuses on earnings forecasts, and over five years you will tend to find that earnings anchor prices. The number is not set in stone, market participants do odd things for long periods of time, we are making assumptions about exit multiples too but this is a good starting point.
Another useful book about expectations implied by prices is Rappaport and Mauboussin’s Expectations Investing.
Learning about valuation is tricky because it requires an understanding of accounting and finance. We have simplified the finance part with our forward earnings rule, now we need to tackle accounting.
We are going to go through a valuation for a fictional housebuilder. The business model is simple, and will allow us to focus on the connection between accounting and valuation. We will discuss how to forecast revenue and costs, cover the pitfalls, and then build a simple model. Two additional sections are included which fill in some detail, I advise reading them but they are supplementary.
How do housebuilders make money? Buy land, convert land into houses, sell houses for more than it cost to buy land and build on it. Let’s begin.
First, we have to think about revenue: the number of houses sold multiplied by the average price.
|Houses sold||Average sale price||Revenue|
Most financial models will focus heavily on revenue. Not all, but when you think about a business, the starting point should be identifying what drives the top-line.
If the company is a website: what is my projection for monthly active users? Can revenue per monthly active user (i.e. price) grow? Can the website sell their ad space in a more targeted way? Are they improving adtech so publishers will pay more for ad space on their website?
Revenue is always simple: volume x price but we have to be precise about the factors behind each item. Is our housebuilder a premium builder? What is the pricing power within that segment? Is our housebuilder regional? What is housing demand and supply like in that region? If they double production, can the market absorb that? How does that interact with pricing?
As you progress further, you may come across the term “mix-shift”. It is rare to see this modelled separately in the UK but is seen in the US. Mix-shift is the change in average price from a change in volume composition as opposed to changing price marks. In our example, we could get a negative mix-shift from higher social housing volume, so we also have to look at the composition of the product portfolio, model these separately if possible, and think about the effect on selling price.
Housebuilding is interesting to forecast because it is a collection of local businesses under one national brand. Aggregates and cement are two similar industries: the value-to-weight ratio of both products are so low that you cannot profitably transport it outside a certain radius so the economics of each site tend to be local. Before canals/railways/supertankers brought down transport costs for many commodities, the economics of most businesses was local. Valuation models should reflect these underlying aspects of the business: if the business is local, our model has to reflect that.
Analysis of these factors has become vastly easier over the last ten years with the proliferation of data about the world. With housebuilders, you can get data from Zoopla about the local housing market, get site-by-site supply, etc. Be totally explicit about how the company is going to outperform though. Without a model, it is easy to read into a situation whatever you want to be true. If you think the company will grow top-line 20%: okay, how? Volume? Price? Be precise.
Next, we move down the income statement, and consider profitability.
|Gross Margin||Operating Margin||Net Margin|
|Gross Profit||Operating Profit||Net Profit|
Costs are divided into three categories. Gross profit is revenue minus costs used directly in the production of revenue, land or materials for example. Operating profit is gross profit minus those costs which cannot be attached to specific revenue, for example head office costs or marketing. Net profit is operating profit minus all other costs, for example taxation.
This is oversimplified. Companies have some latitude to select the category of each cost. Tech companies are a current example: server costs are sometimes included in gross margin, sometimes in operating margin, this makes margin comparisons between companies difficult. And this has a big impact on forecasting as one way to understand costs is by comparing with a similar company.
Costs are more tricky to model than revenue because some costs are variable and scale with revenue, and others are fixed. To model costs accurately you need some business knowledge.
Probably the biggest mistake, one particularly common to sell-side work, is modelling fixed costs as a constant percentage of revenue (i.e. it will, incorrectly, scale with revenue in the model). In some cases, this makes sense. If we are modelling a petrol station: we know that fuel purchasing costs are going to scale with revenue. But when this assumption is incorrect, it leads to significant model errors.
An example of this is Boohoo shortly after their IPO. The company missed estimates, and was panned by analysts for investing heavily in warehouses, which added fixed costs before the company could fill the capacity. What the market missed was that revenue was growing 30% and there was almost no chance the company wouldn’t grow into the capacity: revenue kept growing, they had added the capacity already so profit tripled in two years, and the stock went up 10x. A more subtle example, one that I missed, was Greggs. When a high proportion of revenue growth comes down to the bottom-line, due usually to fixed costs, that is called operating leverage.
A mistake in the opposite direction is to model margins in cyclical industries as trending. In these cases, a company’s margins are profit for another company, so margins tend towards stationarity. Housebuilders are a good example: in 2005/06, housebuilders acquired land aggressively on the basis of historical margins but by 2005, land prices were significantly higher than in the past. Demand fell, and this expensive land depressed gross margins for years as housebuilders worked off that inventory. Today, these companies are more focused on margins.
Related to this, supply changes far more slowly than demand but can have a significant effect in determining long-term industry profits. In our housebuilder example, it wasn’t only demand that fell, there was a huge increase in housing supply too. An improvement in margins will not only lead to pressure from suppliers but attract competition. Margins will, therefore, move with the cycle of capital into the industry and tend towards stationarity over the long-run.
Saying the two previous points another way: companies often claim that they have reached some kind of structural break in margins. In my experience, this is rarely the case. Reductions in cost rarely change competitive dynamics totally: prices fall, suppliers demand more, etc.
A general modelling consideration is that costs also act as an anchor for prices. It is tempting to view costs as something separate to everything else in the model but costs are often related to selling prices. In iron ore, you have the low-cost Australian mines running with economies of scale, and the small Chinese mines running at high-cost. The price of iron ore will usually tend to push on those high-cost suppliers but, clearly, it is unlikely to push on the lowest-cost operations. Not every industry has such huge gaps in production costs though. Google’s prices, for example, are not anchored by cost.
To return to our housebuilder: gross margin is usually a function of labour, materials, and land prices. Of these three, land prices tend to have the most impact at the margin because they tend to rise much faster than selling prices. Despite recent attempts by the industry to grow margins, I would still be inclined to view industry margins as stationary rather than trending. For simplicity, we have modelled constant margins.
So we break each line item down as much as we can, find all the data we can out about those items, forecast revenue and cost, and use these to build a model of the future. We are left with:
|Market Price||Shares Outstanding||Market Cap|
We project 20% volume growth to 2025, 5% price growth, and constant margins. At a market price of 150p, our forecast is for ~12x multiple at the end of our period which isn’t low enough for us. The market price needs to drop below 126p, according to our model.
A full model would break down volume growth by product, by region, and have more detail. Costs would be modelled line by line. But there isn’t enough space to show you what this would look like.
DuPont Formula (Further)
We need to build up some more intuition about financial statements, and the economics of a business as represented by the financials. Businesses vary along two dimensions: margin, and capital intensity. To quantify both, we need the DuPont formula:
Net Margin x Asset Turnover = Return on Assets
We have seen the net margin already but what is asset turnover?
Revenue / Total Assets = Asset Turnover
This is a measure of how much capital is deployed relative to revenue produced. Our housebuilder buys land with cash, has to deploy more cash to buy labour and materials to develop the land, and then gets more cash back when the houses are sold. Asset turnover measures how long that process takes.
We don’t just care about margins – the gap between what houses cost to make and what the company earns from selling them – but also the capital deployed to produce that revenue. Capital is measured on the balance sheet, which is a point-in-time representation of what the company owns, and what it owes.
In our example, if we only had to deploy £10 of assets then that is a very good business. We “turned over” our asset base 10x (100/10) through the year, we “earned” a 10% net margin on every turn so our return on assets was 100% (10*0.1). If we had to deploy £1000 of assets, we earned a 1% return. Not so good.
When you analyse a business, the first thing you want to ask is: margins, and asset turnover. All businesses vary across these two variables, and by understanding these two variables we can make comparisons of businesses across industries.
Tesco, for example, has razor-thin margins. Terrible business? Not so. If you look at the balance sheet, they have negative working capital. This means that suppliers actually pay Tesco to run the business. Tesco only nets 2% on every £1 of revenue but earns this 2% more than 10 times a year. On the opposite side of retail might be an antique dealer turning their inventory over twice every year but making 50% margins. High and low margin businesses can both earn good returns, the DuPont formula explains how this is possible.
However, a business with low returns may still be a good purchase if we are offered the right price. This is beyond the scope of this post but return on equity is linked to valuation through price-to-book. Stephen Penman’s work covers this, and builds intuition between accounting and common valuation measures.
Also, the above is a simplification of DuPont, companies can borrow money. We don’t care about return on assets but return on equity, which is a function of the above formula and leverage. Again, this is beyond the scope of this article but the “real” DuPont formula includes leverage.
Cash Flow (Further)
We haven’t mentioned cash flow because, honestly, accounting is quite boring and cash flow is a simple concept but not particularly easy to understand without knowing about everything else.
Cash flow is a reconciliation between cash on the balance sheet over the accounting period. Our cash balance in 2020 was £100, we started the year at £50, and our cash flow statement tells us where the extra £50 came from over the accounting period. But isn’t cash flow just profit minus dividends and interest costs?
The difference between cash flow and accounting income is accruals, which match the timing of revenue and costs. Accruals are why accounting is complex, and exploiting weaknesses in these rules is how much accounting fraud is executed.
To explain accruals with an example: we start a company with £100 in cash, we build a factory for £100, our factory is magic and produces 200 widgets with no costs with each one selling for £1. What is our profit?
Revenue is £200 with no costs, but what about the £100 we spent on our factory? Do we subtract the factory cost from revenue? There is a mismatch in timing here. Our factory will last for years so we shouldn’t recognise the cost of it all in the current period because if we did our profit next year will jump back to £200, which makes no sense either.
We bridge this gap with a “fake cost” called depreciation. Again, we need to match revenue and costs somehow. Our factory will last, for example, ten years so we recognise the cost of the factory equally over each of those ten years. So our profit in our example is £190 (100 total cost/10 year life = 10, subtract from 200 revenue). We also add an asset on our balance sheet: we initially credit the total cost of the factory showing £100 in assets, then subtract depreciation from this amount as our factory is used. Matching revenue and cost. Matching credit and debit.
Don’t worry if this isn’t totally clear here. As you gain more familiarity with accounts, you will gain intuition. But the point here is that the cash flow statement reconciles the difference between accounting profit and the actual cash flow of the business.
For example, if we look at a company which signs a 10-year contract to maintain a corporate customer’s office, we wouldn’t expect the company to recognise the full value of the contract immediately but over the length of the contract. A large part of Enron’s fraud was recognising revenue for long contracts on signing. Banks did something similar pre-2008 to overstate profitability.
This can work in the other direction too: some companies actually understate accounting profit. An old, but classic, example was British Airways in the late 80s: they owned lots of runways but the depreciation schedule for these runways was much shorter than their actual life, this artificially reduced accounting profit.
The final example is a total grey area: what about a consumer products company that spends heavily on advertising? All of these costs are recognised immediately but advertising can be an investment in the brand, and this spending can accumulate into a huge revenue driver for decades. If rules were changed to allow companies to build advertising assets, it would invite fraud. But the point is that accounting profit doesn’t always recognise underlying economics.
Cash flow is more objective. It is cash-in, cash-out and, depending on the company, you will see cash flow reflect underlying economics over time. Over periods of ten and twenty years, the sum of profit should equal cash flow but over shorter periods, there is an important difference.
5-year forward P/E ratios under 10
Revenue model based on an analysis of volume and price. Cost model considering the difference between fixed and variable costs.
Model should provide an explicit view on how the company hits targets.
Use the DuPont formula to understand and compare business economics
Understand when accounting profit isn’t actually profit, or when cash flow is more representative of the underlying business