The interplay between accounting identities and economic models often remains shrouded in confusion. An accounting identity, by definition, is an equality that must hold true; it’s a tautology. This is designed to categorize and clarify the relationships between different variables. For instance, the identity Assets = Liabilities indicates that whatever value is attributed to “assets,” “liabilities” must match it. This is a definitional truth. It lacks any causal implications and provides no insight into the behavioral relationships among the variables involved.
Moreover, identities are neither testable nor falsifiable; they are always and necessarily true. For example, we cannot test whether an increase in consumption will increase GDP; it will, by definition, do so. If consumption were to decrease GDP, it would imply a mathematical error on our part.
In contrast, economic models are designed to illustrate causal and behavioral relationships. Rather than being established through definitions, models utilize simplifying assumptions and observed behaviors to suggest cause and effect. Consider a basic demand model: Quantity demanded = 10–2P. This equation reflects the causal relationship between price and quantity demanded: as price decreases, quantity demanded increases by 2 units (and vice versa). Unlike an identity, this relationship is not a definition but rather a derived observation that can take various forms depending on the context. Quantity demanded might be quadratic or exhibit other shapes, as it is based on empirical behavior.
Additionally, models are testable and falsifiable. For instance, if we hypothesize that a change in price will influence quantity demanded by 2 units, and our observations reveal that it actually changes by 3 units, our model is invalidated.
The confusion between identities and models likely arises from their similar mathematical appearance. However, to truly grasp their implications, one must discern the differences: a model explains a theoretical causal relationship, while an identity simply describes a definition. Importantly, accounting identities only recount past events.
Michael Pettis, a Carnegie Fellow and finance professor at Peking University, is a notable figure who often misinterprets these concepts, as he frequently writes for large international platforms. Brian Albrecht, chief economist at the International Center for Law & Economics, has penned an insightful blog post that demonstrates how a touch of basic economic reasoning can reveal the flaws, contradictions, and underlying assumptions in Pettis’s arguments. (My co-blogger Scott Sumner has echoed similar critiques.)
Pettis often asserts that the mercantilist policies of other nations, particularly China, compel the United States to maintain a trade deficit. His reasoning stems from the accounting identity that states, in a closed economy, Savings = Investment. Assuming Earth is a closed economy (and let’s face it, we haven’t discovered any sentient alien life just yet), if China boosts its savings, it ostensibly necessitates that another nation must invest more, resulting in a trade deficit. However, accepting Pettis’s argument at face value reveals a fundamental misunderstanding of accounting identities. These identities do not exert some form of coercive influence; rather, they merely track transactions that have already transpired. In essence, they document what has happened, not what will happen. Only actual transactions find their way into the accounts, and accounting serves solely as a record of historical events.
Take the simplified identity Assets = Liabilities. When a firm acquires an asset on credit, both its assets and liabilities increase by the same amount. As the firm repays the debt, both its assets and liabilities decrease equivalently. This must be true because that is the inherent purpose of these categories. Nonetheless, changes in assets and liabilities reflect completed transactions, not future actions. When a transaction occurs, both sides adjust—if the transaction indeed takes place. If it does not, it remains absent from the ledger.
Now, scaling this to the global arena, the accounting identity Investment = Savings – Balance of Trade simply accounts for transactions from the previous time period. If some investment has occurred, it must have been financed through some combination of domestic and foreign savings. However, it does not capture the multitude of investments (and savings) that did not happen because they were unviable at the prevailing interest rates. Because these transactions did not occur, they cannot be recorded, making them invisible to the accounting framework. Again, the accounting identity merely reflects what has transpired, not what is likely to occur.
For another illustration, suppose you have an additional $200,000 and choose to save it. You deposit this sum into a Certificate of Deposit at your local bank. The bank’s assets increase (its reserves grow by your deposit), and its liabilities rise simultaneously (as account holders can withdraw their deposits). While those funds are now available for investment, the deposit itself does not dictate that investments must increase. No one wakes up and says, “I must purchase a house and borrow $200,000 today!” The bank may entice borrowers with interest rates, but no coercion exists, and nothing in the definition necessitates such an outcome. Indeed, if the bank lacks viable borrowers, those funds remain idle in reserves and never contribute to GDP. Thus, while savings may have increased, without corresponding investments, that transaction does not manifest in GDP.
Once we grasp that accounting identities merely document past transactions and lack the power to spur future actions, the foundation of Pettis’s—and indeed, other mercantilist—arguments crumbles. Increased savings in China does not force Americans to incur a trade deficit. Certainly, all else being equal, a rise in savings (increasing the pool of loanable funds) may lower interest rates, thereby boosting the demand for loans, which could elevate U.S. GDP (through enhanced consumption, investment, or government spending) and the trade deficit. Yet, this outcome does not stem from the accounting identity; it arises from fundamental supply-and-demand principles taught in Economics 101. If no one is inclined to borrow or engage in mutually beneficial transactions, then no quantity of increased savings can compel Americans to take on debt.
Â
—
[1] As Brian notes, Pettis might argue that the principles of Economics 101 are irrelevant. However, this is merely another weakness in his reasoning; if you must dismiss established scientific principles to support your case, the case itself is inherently deficient.
