The Economist Numbers Guide

1. The Power of Proportions and Compounding

A percentage increase followed by the same percentage decrease does not leave you back where you started.

The compounding illusion. Business growth, interest, and inflation are governed by the laws of geometric compounding rather than simple addition. When a value changes over multiple periods, we must analyze it using proportions (relative to 1) rather than raw percentages (relative to 100) to avoid costly mathematical errors. For instance, a 50% salary increase followed by a 50% cut actually leaves you with only 75% of your original pay.

Proportions simplify growth. Converting percentages to proportions by shifting the decimal point two places makes complex compounding calculations mentally manageable. This approach is vital for evaluating multi-period changes:

• Compound growth of 6% per year over two years increases a sum by a factor of $1.06 \times 1.06 = 1.1236$.
• A monthly price rise of 2% compounds to an annual inflation rate of 26.8%, not a simple 24%.
• To find a value before a percentage increase, divide the final amount by $(1 + i)$, where $i$ is the increase as a proportion.

Beware of scale. When dealing with large numbers, the terminology of scale can easily mislead. A "billion" in modern business contexts universally represents nine zeros ($10^9$), while scientific notation ($1.25 \times 10^6$) serves as a vital shorthand to prevent errors on calculators and spreadsheets.

2. Demystifying Investment Appraisal (NPV vs. IRR)

An investment is only worthwhile if it promises a higher return than the bank.

Time value of money. Money has a time preference; a dollar today is worth more than a dollar tomorrow because of its earning potential and the risks of inflation and default. Simple interest and payback periods are useful back-of-the-envelope calculations, but they fail to account for the timing of cash flows. To make sound financial decisions, businesses must use discounted cash flow (DCF) techniques to translate future cash flows into today's terms.

NPV versus IRR. Net Present Value (NPV) and Internal Rate of Return (IRR) are the twin pillars of modern investment appraisal, looking at the same financial reality from different angles:

• NPV calculates the net profit of a project in today's money by discounting future cash flows at a chosen interest rate (such as the cost of capital).
• IRR identifies the exact compound interest rate at which the project breaks even, making it easy to compare against alternative investments.
• Sinking funds use the reverse arithmetic to determine how much to save regularly to meet a future capital obligation.

Factoring in reality. Real-world financial analysis must look beyond nominal interest rates to incorporate hidden variables. Effective interest rates must be calculated to standardize different compounding frequencies, and cash flows must be adjusted for inflation and transaction fees to reveal the true economic yield.

3. The Magic of the Normal Distribution

Whenever a measurement is affected by a wide range of small independent factors, observations are distributed in this bell-shaped pattern.

The bell curve. The normal distribution is the most powerful tool in business statistics because it naturally models variables influenced by numerous small, random forces. From the weights of products on an assembly line to the fluctuations of next year's sales, data tends to cluster symmetrically around a central mean. Understanding this pattern allows managers to make highly accurate predictions about risk and performance.

Rules of thumb. A normal distribution is entirely defined by its mean (the midpoint) and its standard deviation (the measure of spread). This structure yields predictable probabilities:

• Approximately 68% of all observations fall within one standard deviation of the mean.
• About 95% of observations are captured within two standard deviations.
• Nearly 100% (99.7%) of the data lies within three standard deviations in either direction.

Quantifying business risk. By converting raw business targets into standardized $z$-scores, managers can calculate the exact probability of various outcomes. For example, a sales manager can determine the precise risk of failing to reach a break-even threshold, turning vague worries into quantifiable probabilities.

4. Decomposing Time Series for Forecasting

The forecasting trick is to decompose a series into the trend, cycles, seasonality and residual, extend each into the future, and then recombine them into one again.

Decomposing time series. Business data recorded over time is rarely a simple, straight line; it is a complex composite of multiple overlapping forces. To build reliable forecasts, managers must break down a historical time series into its constituent parts, project each component individually, and then assemble them back together. This systematic approach prevents short-term seasonal spikes from being mistaken for long-term structural growth.

Four core components. Every business time series can be decomposed into four distinct elements that explain its behavior:

• The Trend: The long-run, underlying direction of the data (upward, downward, or flat).
• The Cycle: Medium-to-long-term rhythmic fluctuations driven by broader economic or industry business cycles.
• Seasonality: Short-term patterns that repeat reliably every twelve months, such as holiday shopping surges.
• The Residual: The random, unexplained noise left over after accounting for the other three components.

Smoothing and adjusting. Moving averages and exponential smoothing are invaluable tools for stripping away random noise to reveal the true underlying trend. Once the seasonal adjustment factors are calculated, managers can evaluate monthly performance on a seasonally adjusted basis, comparing like with like.

5. Identifying Relationships with Regression Analysis

The coefficient of determination, better known as r2, indicates how much of the change in y is explained by x.

Causal forecasting power. While extrapolation assumes the future will simply mirror the past, causal forecasting seeks to identify the underlying drivers of business performance. Regression analysis is the mathematical engine used to find the line of best fit between a dependent variable (like sales) and one or more independent variables (like advertising spend or local population). This allows businesses to move from guessing to calculating the impact of their strategic decisions.

Evaluating relationship strength. Running a regression on a spreadsheet will always yield an equation, but managers must look at specific statistical indicators to judge its reliability:

• The correlation coefficient ($r$) measures the strength and direction of the linear relationship between $-1$ and $+1$.
• The coefficient of determination ($r^2$) reveals the percentage of variation in the dependent variable explained by the model.
• The standard error of the estimate provides a standard deviation for the forecast, establishing a range of probable outcomes.

Avoiding analytical traps. Managers must remember that correlation does not equal causation; two unrelated variables can appear highly correlated due to a shared relationship with a third factor, like inflation. Furthermore, regression models must be checked for collinearity among independent variables and autocorrelation among residuals to ensure the forecast is statistically sound.

6. Inference, Sampling, and Hypothesis Testing

The key hypothesis is always formed cautiously in favour of no change or no action.

Inference from samples. In business, collecting complete data on an entire population is often too expensive, time-consuming, or physically impossible. Sampling allows managers to draw highly accurate conclusions about a population by examining a small, representative subset. The central limit theorem guarantees that as sample sizes grow past 30, the sample means will be normally distributed, enabling precise estimation of confidence intervals.

Rigorous hypothesis testing. Hypothesis testing provides a structured framework for making decisions based on sample data, forcing managers to state their assumptions clearly:

• The Null Hypothesis ($H_0$) is the conservative baseline, assuming no change, no effect, or no difference.
• The Alternative Hypothesis ($H_a$) is the claim the manager hopes to prove, requiring strong evidence to accept.
• Type I Error ($\alpha$) occurs when we incorrectly reject a true null hypothesis (an error of commission).
• Type II Error ($\beta$) occurs when we fail to reject a false null hypothesis (an error of omission).

Balancing decision risks. Just as a pedestrian would rather wait unnecessarily than step into traffic, businesses must balance the costs of Type I and Type II errors. By setting strict significance levels, organizations can control their exposure to risk, ensuring they only implement changes when the statistical evidence is overwhelming.

7. Structuring Decisions Under Uncertainty and Risk

For any decision, the expected value of perfect information evpi = eppi – ep.

Structuring the unknown. When faced with complete uncertainty, managers can use decision tables to map out their options against uncontrollable states of nature. By assigning payoffs to every possible outcome, the decision-making process becomes transparent and logical. Even without probabilities, techniques like maximax (for optimists) and maximin (for pessimists) provide clear frameworks for action.

Quantifying risk and value. Once probabilities can be assigned to future scenarios, managers can calculate the Expected Payoff (EP) for each decision alternative:

• Expected Payoff is the weighted average of all possible outcomes, serving as a long-run guide for repeated decisions.
• The Expected Value of Perfect Information (EVPI) sets the absolute maximum budget a company should spend on market research.
• The Expected Value of Sample Information (EVSI) adjusts this limit to account for the realistic, imperfect nature of actual research.

Incorporating utility. Standard financial calculations assume that every dollar has the same value, but human decision-makers are rarely risk-neutral. Utility theory allows managers to factor their personal or corporate risk tolerance into the decision tree, converting monetary payoffs into subjective utility scores to find the option that truly aligns with their comfort level.

8. Optimizing Operations via Queueing and Stock Control

If 95 customers arrive every hour and a single cashier can serve 100 an hour, the average queue will contain 18 irritated people.

The cost of waiting. Queueing theory is a vital operational tool for balancing the cost of idle service capacity against the cost of customer dissatisfaction. It reveals a counterintuitive truth: even when service capacity exceeds the average arrival rate, random fluctuations will still create significant bottlenecks. Understanding this math prevents managers from understaffing service points and losing valuable customer goodwill.

Operational queueing variables. Analyzing a queueing system requires identifying a few key parameters that define how customers move through the system:

• Arrival Rate ($\lambda$): The frequency of incoming customers, typically modeled using a Poisson distribution.
• Service Rate ($\mu$): The speed at which a single channel can process customers, often following an exponential distribution.
• Utilization Rate ($\lambda/\mu$): The percentage of capacity at which the system runs, which must remain below 1 to prevent infinite queues.

Balancing inventory costs. Stock control applies a similar balancing logic to physical goods, weighing the costs of ordering against the costs of carrying inventory. The Economic Order Quantity (EOQ) formula identifies the optimal order size that minimizes these competing costs, while safety stocks are calculated using normal distributions to protect against unexpected surges in demand.

9. Strategic Gaming and Markov Chains

In isolation, they all decide rationally on the first option. As a result, all receive tougher sentences than if they all kept silent.

Strategic game theory. Business is rarely played in a vacuum; the success of your decisions depends heavily on the active, competitive responses of your rivals. Game strategy provides a mathematical framework for analyzing these competitive dynamics, helping managers identify optimal moves in pricing wars, product launches, and market entry. It reveals that purely self-interested, rational actions can sometimes lead to mutually destructive outcomes.

Core gaming concepts. To navigate competitive situations, managers must understand the structural rules of different types of games:

• Zero-Sum Games: Situations where one player's gain is directly equal to the other player's loss.
• Saddle Point: A stable equilibrium where both players' optimal strategies intersect, determining the value of the game.
• Mixed Strategy: A randomized approach used when no single move is dominant, keeping competitors guessing.

Predicting system transitions. While game theory models competitive moves, Markov chains model systems that transition from one state to another over time. By constructing a transition matrix of probabilities, businesses can predict long-term market shares, customer loyalty patterns, and the likelihood of accounts receivable turning into bad debts as the system settles into equilibrium.

10. Linear Programming and Network Optimization

The corner points rule holds good only when the boundary to the set of feasible solutions is convex.

Optimizing under constraints. Linear programming is a powerful mathematical technique for finding the absolute best way to allocate limited resources to achieve a specific goal, such as maximizing profit or minimizing cost. Whether determining the optimal product mix for a factory or the most cost-effective delivery routes, this approach guarantees a mathematically perfect solution within the boundaries of your real-world constraints.

The geometry of optimization. For simple problems with two or three variables, linear programming can be visualized geometrically on a graph:

• The constraints draw boundary lines that define a shaded "feasible region" of all possible solutions.
• The optimal solution will always lie at one of the extreme corner points of this feasible region.
• Dual formulations of the problem reveal "shadow prices," which calculate the exact value of acquiring one additional unit of a constrained resource.

Solving network paths. Network optimization extends these programming principles to physical systems, solving complex routing problems. Managers can use simple algorithms to identify the shortest path for deliveries, the minimum span to connect multiple utility points, or the maximum flow of goods through a distribution network, driving massive efficiency gains.