Proxy: How To Know Everything, But Not Really

A proxy is a variable that is not in itself directly relevant, but that serves in place of an unobservable or immeasurable variable. Per-capita GDP is a proxy of a standard of living. Birth country is a proxy for race. Body mass index is a proxy of health. Proxies are readily observable and easily measurable variables useful as stand-ins when measuring the target variable is impractical, costly, or impossible. When we cannot measure something directly, we use a proxy.

A common proxy you are probably aware of is the measurement of tree rings to see how old trees and forests are. The science is called dendroclimatology. Using tree rings, scientists have estimated many local climates for hundreds of years. Sometimes other climate proxy records are combined with tree-ring studies and past regional and global climates have been estimated with relative accuracy. Similarly, paleoclimatologists use fossils to reconstruct past climates. Other ways proxies find their use in climate observations: coral rings and bands, pollen grains, lake and ocean sediments, and most commonly, ice cores.

They are also used widely in statistics to form correlations to variables of interest. GDP per capita to assess quality of life, tree ring width to measure historical climate, years of education and GPA for intelligence, and height and/or muscle growth for hormone levels. The idea remains the same. A proxy is used to approximate the target variable without touching the variable itself. They are often easier and more practicable to observe than the primary target. The effectiveness depends on how strongly it correlates to the actual subject, so over-reliance can result in a distortions if a proxy fails to capture the full complexity of the variable at hand.

Test scores for student knowledge, step counts for physical activity levels, and GDP for national well-being are only as dependable as the individual, team, or phenomenon you are studying. Thus the benefits are clear: simplicity, making complex phenomena more manageable; feasibility, since direct measurement is not always possible or practical; and speed, because a proxy relies on accessible stand-ins to help you decide. The risks are as well: misalignment, since using the wrong proxy results in misleading outcomes; perverse incentive, because overemphasizing proxies can make you behave in a way that undermines the true goal; and oversimplification, since at the core, proxies are vehicles that strip away nuance and may fail to account for broader contexts or standout cases.

Real life implications of proxies:

• Business: you might use monthly revenue as a proxy for business health, and this is okay, but supplement it with other indicators like customer retention to avoid misrepresenting business stability;

• Healthcare: body mass index is useful for overall health, but you have to pair it with other measures like cholesterol levels, fat-free mass ratio, and physical fitness to get a more comprehensive health picture;

• Environment: carbon dioxide levels are a relatively reliable proxy for climate change progression, but solely tracking carbon dioxide without other data such as ocean temperatures and biodiversity will not capture the broader impact;

• Education: exam grades are a proxy for how well the students understand the subject, but assessments like projects and oral presentations can give you a more well-rounded view of student competence;

• Economics: GDP is a proxy for national prosperity, yet you can augment it with metrics like the Gini coefficient for income inequality or the human development index for a nuanced understanding;

• Personal: hours spent practicing a skill is a proxy for improvement, only they do not indicate actual performance such as accuracy or speed, so take relevant measures into account to ensure meaningful progress;

• Fitness: heart rate is a useful proxy for workout intensity, so calibrate those cardio zones based on your fitness goals, lower for fat burn, higher for endurance, and so on;

• Investments: stock price is a proxy for company value, but supplement it by analyzing fundamentals like earnings reports to avoid speculative risk-taking;

• Relationships: frequency of communication is a proxy for how strong a relationship is, yet it’s often skewed because it fails to take into account variables like emotional depth and quality time spent together.

How you might employ proxies as a thinking tool: (1) clarify the objective, identifying the variable you want to measure or act upon, like customer loyalty using repeat purchases as a proxy; (2) ensure that the proxy strongly correlates or accurately represents the target variable, and customers coming back definitely hits the mark; (3) recognize limitations, understanding what the proxy measures and what it does not, because the customer might just be incentivized by a sale or recent marketing campaign you launched and not because they are loyal to your brand; (4) avoid over-reliance, use proxies as tools, not substitutes for understanding the actual phenomenon; (5) iterate, reassess, and refine your measurements because proxies are already estimates, making them sensitive to changing conditions.

Thought-provoking insights. “Not everything that can be counted counts, and not everything that counts can be counted.” attributed to Albert Einstein, it highlights the limitations of proxies in representing the full complexity of things. The parable of the Drunkard’s Search: the drunk searches for their keys under a streetlight, not because he lost them there, but because the light is better; similarly, proxies can skew your vision because they are easier to measure, not because they are the most relevant or useful. “You manage what you measure.” emphasizes how proxies can shape your priorities and actions, for better or worse. Use them to serve the ultimate objective. Navigate complexity when you don’t have time using proxies.

Questions to reflect on:

1. How can I use a proxy to make informed decisions when direct information is unavailable?

2. What potential risks can I run into if I rely on proxies too much?

3. In what situations have I successfully used proxies to achieve my goals? Where have they not worked?

4. What strategies can I use to combine multiple proxies for a comprehensive analysis?

5. How do proxies influence my perception and interpretation of data?

Quotes to make your brain break a sweat:

1. "The best way to predict the future is to create it." - Peter Drucker, management consultant and author.

2. "In God we trust; all others must bring data." - W. Edwards Deming, American engineer and statistician.

3. "The map is not the territory." - Alfred Korzybski, Polish-American philosopher and scientist.

4. "You can't manage what you can't measure." - Peter Drucker.

Example use cases:

1. Investing: investors use proxies to evaluate the potential of a company when direct financial data is not available. For instance, Charlie Munger used the concept to identify companies that might ride the wave of recent technological advancements. This led him to invest in Microsoft when it was still a startup.

2. Real estate: agents use proxies like neighborhood crime rates and school district quality to estimate property values when direct comparisons are not possible or clear.

3. Marketing: marketers use social media engagement as a proxy for customer satisfaction and brand loyalty, helping them gauge public perception and adjust their campaigns accordingly.

4. Education: teaching faculty use standardized test scores as proxies for student learning and academic performance, although they recognize how limited and potentially biased these measures are.