A brief account of how I studied for my PhD qualifying exam, and what turned out to matter most.
I’d like to share how I studied for my PhD qualifying exam, perhaps to help others.
My most important realization while studying was this: it is necessary to learn derivation rather than memorization. A superficial understanding of a topic is not enough; what is expected of you in the exam is not remembering the result, but being able to figure out how that result was reached.
For example, you have not truly learned the concept unless you can derive the relationship from utility to risk premium and then to ARA.
Therefore, I studied each topic at three levels:
Without a combination of these three, the information is not truly internalized.
During my study process, I also realized that textbooks teach, but do not impart the exam logic; that is why past exam questions are so important, if you can find them. Past questions teach you which perspective to think from.
Also, I never relied on the feeling of “I understood.” To test myself, I would close the paper and try to rewrite and derive the topic from scratch; if I could not, it meant I had not actually learned it.
The biggest mistake was passive studying: reading, underlining, and similar habits. Real learning comes from writing, deriving ideas, and solving problems.
Learning ≠ solving.
In a PhD qualifying exam, what matters is not just knowing the information, but being able to solve the problem, build it from scratch, and make connections. For example, you might know about asymmetric information, but you may be asked to explain it at the market microstructure level. Therefore, it is crucial to make the connection well, and understanding the fundamentals is key.
The best way to know if you have truly learned a subject is to be able to explain it to someone else in simple, straightforward language. If you cannot explain it, or if the other person does not fully understand, then you have not fully understood it yourself.
This process taught me this: what makes the difference is not working hard, but working correctly.
When studying banking, the most critical thing is not memorizing institutions, but understanding contract design. I viewed banks not as “institutions,” but as a collection of contracts. Because a bank fundamentally designs deposit contracts and loan contracts; the real question is why these contracts are structured this way.
I also addressed the essence of financial intermediation within this framework: optimal contract design under information asymmetry. Markets alone are insufficient because problems like adverse selection and moral hazard exist; banks solve these problems through contracts.
Deposit contracts, in particular, are central to this. Questions like why demand deposits are offered and why early pickers win go to the heart of banking theory. The Diamond–Dybvig model shows that providing liquidity and the risk of bank runs essentially stem from the same contract.
In loan contracts, I came to see that lending is not just about giving money; it is actually a process of screening, monitoring, and incentive design. The optimal contract should be designed to ensure that the borrower behaves correctly.
My perspective on risk issues also changed during this process. Instead of memorizing terms like interest rate risk, liquidity risk, and credit risk individually, I focused on understanding how these risks are distributed within contracts. Because risk is not just something that is measured, but an element that is redistributed within a contract.
More advanced topics such as securitization, off-balance sheet transactions, and shadow banking are extensions of the same logic: repackaging risk and rewriting the contract.
The biggest mistake I saw during this process was studying banking through products. However, the correct approach always involves these two questions: what problem exists, and which contract solves this problem?
Ultimately, learning banking is actually learning contract theory; everything else is detail.
Corporate finance is not about memorizing financial decisions, but about understanding incentive design. Although companies’ capital structure, dividend policies, or investment decisions may appear as financial choices on the surface, they are fundamentally decisions made under information asymmetry and agency problems.
Therefore, I did not study corporate finance through the question “what does each company do?” Instead, I focused on the question: what is the incentive problem, and what mechanism solves it?
I used the Modigliani–Miller framework as a starting point. Because MM is not actually a model of the real world, but a benchmark: if there are no frictions, the financial structure is irrelevant. Therefore, the important thing is to understand why MM does not work.
Capital structure begins to make sense when taxes, bankruptcy costs, agency problems, and information asymmetry come into play. This perspective connected all the issues.
In the agency theory part, I viewed the firm as a collection of contracts. The conflict of interest between the manager and the shareholder, the free cash flow problem, and control issues are actually incentive alignment problems.
Jensen–Meckling and the subsequent literature showed me that financial decisions are also a control and incentive mechanism. Even the use of debt is not only financing, but also a tool for disciplining the manager.
On the asymmetric information side, I focused on signaling and adverse selection mechanisms. Models like Myers–Majluf taught me that firms’ financing choices (for example, internal financing > debt > equity) actually stem from their information structure. In other words, financial structure is a signal given to the market.
At this point, instead of memorizing capital structure theories, I considered them all through the same problem: who holds the information, and how is this information priced?
More advanced topics such as IPOs, dividend policy, corporate structure, and institutions were extensions of the same logic. IPO underpricing, payout policy, or corporate governance differences actually reflect the same fundamental problem that arises in different contexts: information, incentives, and control.
Therefore, I studied the topics not separately, but as parts of a single framework.
The biggest mistake I saw in this process was studying corporate finance like a “list of theories.” However, the correct approach always comes back to the question: what is the problem, and what mechanism solves this problem?
Ultimately, learning corporate finance is about understanding not the financial decisions of firms, but the incentives and information structure behind those decisions.
Econometrics is not really about learning techniques, but about understanding where estimators come from and what they measure. I took econometrics as two separate courses, one empirical and one theoretical, but here I focus on the theoretical part. Because empirical applications without understanding the theory often remain mechanical.
Therefore, while studying, I focused more on “why is this estimator like this?” rather than “how do I apply it?” I especially based my theoretical foundation on Greene’s Econometric Analysis (the green book) and worked by deriving each result.
For OLS, clarifying the distinction between β, β̂, and β̃ was critical: β represents the actual population parameter, β̂ the estimate obtained from the sample, and β̃ the alternative estimators that emerge under different assumptions.
Analyses that fail to understand this distinction miss the point of what is actually being estimated. Therefore, I studied not only the result, but also in detail how OLS is derived: the minimization problem, normal equations, projection logic, and the assumptions under which it satisfies particular properties.
For me, the Gauss–Markov theorem was not just a result to memorize, but a chain of logic: under what assumptions is β̂ the best linear unbiased estimator (BLUE), and what does “best” mean?
Similarly, I approached topics like heteroskedasticity and serial correlation not as a “list of problems,” but by understanding how the distribution of β̂ changes when these assumptions are violated. Because the real issue is not knowing the tests, but understanding the estimator’s behavior.
Asset pricing models are essentially about understanding how risk is priced. Therefore, I studied topics like CAPM, mean–variance analysis, and APT not as separate theories, but as different statements of the same problem.
The fundamental question was always this: which risks are priced, and why do some risks yield a premium while others do not?
My starting point was mean–variance analysis. Because this framework simplifies investor behavior by showing that investors strike a balance between expected return and risk (variance). The efficient frontier and portfolio selection actually form the basis of all subsequent asset pricing models. It is impossible to understand CAPM without first understanding this framework.
Markowitz’s 1952 paper and Sharpe’s 1964 paper are quite important in this regard.
I studied CAPM not as a formula, but as a result of equilibrium. The model states that the priced risk is not individual risk, but systematic risk. That is, what matters is not the volatility of an asset, but how it moves with the market (β).
At this point, it became critical to work through CAPM by deriving it. I focused on understanding the process starting from mean–variance optimization, moving to the market portfolio, and then to the security market line. Thus, CAPM became not an equation to be memorized, but a chain of logic for me.
Roll’s critique (1977), which found CAPM insufficient, is very important at this point. Multifactor models such as the Fama–French and Fama–MacBeth papers emerged in this broader context.
I considered APT and multifactor models within the same framework. While CAPM works with a single risk factor (market risk), APT showed me that if there is no arbitrage, multiple systematic risk factors can also be priced.
So, even if the model changes, the logic remains the same: risk → pricing. Therefore, I saw APT not as an alternative to CAPM, but as a generalization of it.
Ultimately, learning asset pricing is not about knowing the formulas; it is about understanding the relationship between risk and return.
Behavioral finance is not simply about saying “people behave irrationally”; the real issue is whether this irrationality is systematic and how it is reflected in prices. Classical financial theory assumes investors are rational, but behavioral finance explains why and how this assumption breaks down.
Therefore, I approached the subject not as a list of biases, but by understanding the mechanisms by which deviations from the rational model occur.
My starting point was efficient markets and limits to arbitrage. Because irrationality alone is not enough; what matters is why these errors are not immediately eliminated by arbitrage. The Shleifer–Vishny framework showed me that if arbitrage is not limitless, irrational behavior can permanently affect prices. This perspective elevates behavioral finance from a mere “list of anomalies” to a theoretical framework.
Instead of memorizing biases individually, I considered them under specific groups. Prospect theory became central to this framework: loss aversion, reference dependence, and the concave/convex value function differentiate investor behavior from classical expected utility.
In addition, heuristics like overconfidence, representativeness, and availability cause investors to make systematic errors. The important thing was not just knowing that these biases exist, but understanding which errors occur under which circumstances.
The reflection of these biases in markets is a separate layer. Findings such as overreaction, underreaction, momentum, and the disposition effect are actually aggregations of individual errors. For example, investors selling winners early and holding onto losers, or making too many trades due to overconfidence, directly affect price dynamics.
In more advanced areas such as sentiment, attention, and emotion, my approach remained the same: I tried to understand how these factors influence prices. Baker–Wurgler sentiment measures or attention-based models showed me that prices are shaped not only by risk, but also by psychological states.
The biggest mistake I saw in this process was treating behavioral finance as a “list of irrational behaviors.” The correct approach, however, boils down to this question: what bias exists, what behavior does this bias produce, and what market outcome does this behavior lead to?
In other words, learning behavioral finance is not about memorizing mistakes; it is about understanding how these mistakes systematically translate into price changes.