What We’re Reading (Week Ending 27 August 2023) - 27 Aug 2023
Reading helps us learn about the world and it is a really important aspect of investing. The legendary Charlie Munger even goes so far as to say that “I don’t think you can get to be a really good investor over a broad range without doing a massive amount of reading.” We (the co-founders of Compounder Fund) read widely across a range of topics, including investing, business, technology, and the world in general. We want to regularly share the best articles we’ve come across recently. Here they are (for the week ending 27 August 2023):
1. Why Lehman Brothers Failed When It Did – Joe Pimbley
In 2008, security firms operated with high leverage and significant amounts of short-term debt. Lehman had $26 billion of equity supporting $639 billion of assets and its high leverage was not unusual among security firms. But at that ratio, a 4% decline in assets wipes out equity. Meanwhile, reliance on the continuous rolling of short-term debt requires the security firm to always maintain lender confidence. Lenders’ perception of solvency becomes more important than the actual fact of solvency.
When the highly leveraged, short-term debt, security firm business model met the asset-value destruction of the Great Financial Crisis, Lehman was not the only security firm to fail. All major US firms failed to one degree or another. Besides Lehman’s outright bankruptcy, Bear Stearns and Merrill Lynch were merged into commercial banks. I believe Goldman Sachs and Morgan Stanley would have defaulted on their short-term borrowings had the Fed not permitted them to convert to bank holding companies and gain access to discount window liquidity…
…A place to begin chronicling factors specific to Lehman’s failure is the beginning of 2006. That was when the firm’s management decided to make more long-term investments.[2] Rather than remaining focused on security distribution and brokerage, Lehman increased its own holdings in commercial real estate, leveraged loans, and private equity. In our report to the bankruptcy court, we described this strategic change as a shift from the “moving business” to the “storage business.”
One year later in early 2007, Lehman management viewed the incipient financial crisis as an opportunity for the firm to gain market share and revenue from competitors that were retrenching and lowering their risk profiles. Lehman did not think the subprime mortgage crisis would spread to the general economy or even to its growing commercial real estate portfolio. Lehman had boldly taken on assets and assumed risk in the 2001-02 economic downturn. Its risk-taking back then had paid off and it hoped such contrarian boldness would again prove profitable.
Lehman’s pace of principal investments in commercial real estate, leveraged loans, and private equity increased in the first half of 2007 as other security firms reduced risk and hunkered down. It committed $11 billion to acquire Archstone REIT in May 2007 and ended up funding the riskiest $6 billion of that in October when it couldn’t find enough buyers to take it out of its commitment. Other bridge loans and bridge equity positions also became similarly stuck on its balance sheet. Its mortgage subsidiaries were slow to stop making residential mortgage loans and Lehman ended up holding mortgage-backed bonds and mortgage-bond-backed collateralized debt obligations it couldn’t sell.
To take on these risky assets, Lehman’s management raised all its internal risk limits: firm-wide, line-of-business, and even single-name risk limits. Or they ignored the limits they had set. Management was not fulsome in its disclosures to its board of directors about the risks it assumed and Lehman’s board did not press management for important information. In theory, Lehman’s compensation policy penalized excessive risk taking, but in practice it rewarded employees on revenue with minimal attention to associated risk.
Not only were these investments risky from the perspective of potential market value losses; they were risky from the point of view of financing. By their nature, real estate, leveraged loans, and private equity are hard to value and less liquid. It is difficult to determine how quickly and how severely they could lose value. These characteristics mean the ability to finance these assets cannot be assumed. If lenders worry about the realizable value of assets offered as loan security, they will lower the amount they will lend against those assets or cease lending against them altogether. Most of Lehman’s secured debt had overnight tenors, so lenders could stop rolling over their loans to Lehman on any business day!
Lehman’s management only began to cut back on leveraged loan acquisitions in August 2007 and they waited until later in 2007 to cut back on commercial real estate purchases. Yet deals in the pipeline caused Lehman’s assets to grow $95 billion to $786 billion over the quarter ending February 2008. The firm did not begin to sell assets in earnest until March 2008, but only got assets down to $639 billion by May 2008.
Lehman’s management deliberately deceived the world about the firm’s financial condition. Management used an accounting trick to temporarily remove $50 billion of assets from the firm’s balance sheet at the end of the first and second quarters of 2008. In so-called “repo 105” transactions, Lehman pledged assets valued at 105% or more of the cash it received. Relying on a legal opinion from a UK law firm addressing English law, Lehman deducted the assets from its balance sheet. No other security firm used this stratagem in 2008 and Lehman did not disclose its use.
Lehman’s management touted the firm’s “liquidity pool,” the sum of cash and assets readily convertible into cash and as late as two days before bankruptcy claimed this pool equaled $41 billion. In fact, only $2 billion of those assets were readily monetizable.
From January to May 2008, while its competitors raised equity, Lehman did not. Lehman’s management rejected offers from interested investors because they did not want to issue equity at a discount to market price. Management thought doing so would make the firm seem vulnerable. Lehman did not issue common stock in 2008 until a $4 billion issuance in June.
2. China’s 40-Year Boom Is Over. What Comes Next? – Lingling Wei and Stella Yifan Xie
For decades, China powered its economy by investing in factories, skyscrapers and roads. The model sparked an extraordinary period of growth that lifted China out of poverty and turned it into a global giant whose export prowess washed across the globe.
Now the model is broken.
What worked when China was playing catch-up makes less sense now that the country is drowning in debt and running out of things to build. Parts of China are saddled with under-used bridges and airports. Millions of apartments are unoccupied. Returns on investment have sharply declined.
Signs of trouble extend beyond China’s dismal economic data to distant provinces, including Yunnan in the southwest, which recently said it would spend millions of dollars to build a new Covid-19 quarantine facility, nearly the size of three football fields, despite China having ended its “zero-Covid” policy months ago, and long after the world moved on from the pandemic…
…What will the future look like? The International Monetary Fund puts China’s GDP growth at below 4% in the coming years, less than half of its tally for most of the past four decades. Capital Economics, a London-based research firm, figures China’s trend growth has slowed to 3% from 5% in 2019, and will fall to around 2% in 2030.
At those rates, China would fail to meet the objective set by President Xi Jinping in 2020 of doubling the economy’s size by 2035. That would make it harder for China to graduate from the ranks of middle-income emerging markets and could mean that China never overtakes the U.S. as the world’s largest economy, its longstanding ambition.
Many previous predictions of China’s economic undoing have missed the mark. China’s burgeoning electric-vehicle and renewable energy industries are reminders of its capacity to dominate markets. Tensions with the U.S. could galvanize China to accelerate innovations in technologies such as artificial intelligence and semiconductors, unlocking new avenues of growth. And Beijing still has levers to pull to stimulate growth if it chooses, such as by expanding fiscal spending.
Even so, economists widely believe that China has entered a more challenging period, in which previous methods of boosting growth yield diminishing returns…
…The transition marks a stunning change. China consistently defied economic cycles in the four decades since Deng Xiaoping started an era of “reform and opening” in 1978, embracing market forces and opening China to the West, in particular through international trade and investment.
During that period, China increased per capita income 25-fold and lifted more than 800 million Chinese people out of poverty, according to the World Bank—more than 70% of the total poverty reduction in the world. China evolved from a nation racked by famine into the world’s second-largest economy, and America’s greatest competitor for leadership.
Academics were so enthralled by China’s rise that some referred to a “Chinese Century,” with China dominating the world economy and politics, similar to how the 20th century was known as the “American Century.”
China’s boom was underpinned by unusually high levels of domestic investment in infrastructure and other hard assets, which accounted for about 44% of GDP each year on average between 2008 and 2021. That compared with a global average of 25% and around 20% in the U.S., according to World Bank data.
Such heavy spending was made possible in part by a system of “financial repression” in which state banks set deposit rates low, which meant they could raise funds inexpensively and fund building projects. China added tens of thousands of miles of highways, hundreds of airports, and the world’s largest network of high-speed trains.
Over time, however, evidence of overbuilding became apparent.
About one-fifth of apartments in urban China, or at least 130 million units, were estimated to be unoccupied in 2018, the latest data available, according to a study by China’s Southwestern University of Finance and Economics…
…Guizhou, one of the poorest provinces in the country with GDP per capita of less than $7,200 last year, boasts more than 1,700 bridges and 11 airports, more than the total number of airports in China’s top four cities. The province had an estimated $388 billion in outstanding debt at the end of 2022, and in April had to ask for aid from the central government to shore up its finances.
Kenneth Rogoff, a professor of economics at Harvard University, said China’s economic ascent draws parallels to what many other Asian economies went through during their periods of rapid urbanization, as well as what European countries such as Germany experienced after World War II, when major investments in infrastructure boosted growth.
At the same time, decades of overbuilding in China resembles Japan’s infrastructure construction boom in the late 1980s and 1990s, which led to overinvestment.
The solution for many parts of the country has been to keep borrowing and building. Total debt, including that held by various levels of government and state-owned companies, climbed to nearly 300% of China’s GDP as of 2022, surpassing U.S. levels and up from less than 200% in 2012, according to Bank for International Settlements data.
Much of the debt was incurred by cities. Limited by Beijing in their ability to borrow directly to fund projects, they turned to off-balance sheet financing vehicles whose debts are expected to reach more than $9 trillion this year, according to the IMF.
Rhodium Group, a New York-based economic research firm, estimates that only about 20% of financing firms used by local governments to fund projects have enough cash reserves to meet their short-term debt obligations, including bonds owned by domestic and foreign investors…
…In Beijing’s corridors of power, senior officials have recognized that the growth model of past decades has reached its limits. In a blunt speech to a new generation of party leaders last year, Xi took aim at officials for relying on borrowing for construction to expand economic activities…
…The most obvious solution, economists say, would be for China to shift toward promoting consumer spending and service industries, which would help create a more balanced economy that more resembles those of the U.S. and Western Europe. Household consumption makes up only about 38% of GDP in China, relatively unchanged in recent years, compared with around 68% in the U.S., according to the World Bank.
Changing that would require China’s government to undertake measures aimed at encouraging people to spend more and save less. That could include expanding China’s relatively meager social safety net with greater health and unemployment benefits.
Xi and some of his lieutenants remain suspicious of U.S.-style consumption, which they see as wasteful at a time when China’s focus should be on bolstering its industrial capabilities and girding for potential conflict with the West, people with knowledge of Beijing’s decision-making say.
The leadership also worries that empowering individuals to make more decisions over how they spend their money could undermine state authority, without generating the kind of growth Beijing desires.
A plan announced in late July to promote consumption was criticized by economists both in and outside China for lacking details. It suggested promoting sports and cultural events, and pushed for building more convenience stores in rural areas.
Instead, guided by a desire to strengthen political control, Xi’s leadership has doubled down on state intervention to make China an even bigger industrial power, strong in government-favored industries such as semiconductors, EVs and AI.
While foreign experts don’t doubt China can make headway in these areas, they alone aren’t enough to lift up the entire economy or create enough jobs for the millions of college graduates entering the workforce, economists say.
3. LTCM: 25 Years On – Marc Rubinstein
To understand, it helps to model LTCM not as a hedge fund but as a bank (although it’s also true that the best model for a bank is often a hedge fund). Roger Lowenstein, author of When Genius Failed, acknowledges as much in the subtitle of his book: “The Rise and Fall of Long-Term Capital Management: How One Small Bank Created a Trillion-Dollar Hole.”
The model reflects LTCM’s heritage. John Meriwether ran the arbitrage desk at Salomon Brothers becoming vice chair of the whole firm, in charge of its worldwide Fixed Income Trading, Fixed Income Arbitrage and Foreign Exchange businesses. In the years 1990 to 1992, proprietary trading accounted for more than 100% of the firm’s total pre-tax profit, generating an average $1 billion a year. LTCM was in some ways a spin-off of this business.
Indeed, LTCM partners viewed their main competitors as the trading desks of large Wall Street firms rather than traditional hedge funds. Thus, although they structured their firm as a hedge fund (2% management fee, 25% performance fee, high watermark etc) they did everything they could to replicate the structure of a bank. So investors were required to lock-up capital initially for three years to replicate the permanent equity financing of a bank (hence “Long-Term Capital Management”). They obtained $230 million of unsecured term loans and negotiated a $700 million unsecured revolving line of credit from a syndicate of banks. They chose to finance positions over 6-12 months rather than roll financing daily, even at the cost of less favourable rates. And they insisted that banks collateralise their obligations to the fund via a “two way mark-to-market”: As market prices moved in favour of LTCM, collateral such as government bonds would flow from their counterparty to them.
If there was one risk LTCM partners were cognisant of it is that they might suffer a liquidity crisis and not be able to fund their trades. It was a risk they took every effort to mitigate.
But in modelling themselves as a bank, they forgot one key attribute: diversification.
“We set up Long-Term to look exactly like Salomon,” explains Eric Rosenfeld. “Same size, same scope, same types of trades… But what we missed was that there’s a big difference between the two: Long-Term is a monoline hedge fund and Salomon is a lot of different businesses – they got internal diversification from their other business lines during this crisis so therefore they could afford to have taken on more risk. We should have run this at a lower risk.”
It’s a risk monolines in financial services often miss. And LTCM wasn’t the only monoline to fall victim to market conditions in 1998. In the two years that followed, eight of the top 10 subprime monolines in the US declared bankruptcy, ceased operations or sold out to stronger firms. The experience prompted some financial institutions – such as Capital One – to embrace a more diversified model.
When the global financial crisis hit in 2007, monoline firms went down first. And in the recent banking crisis of 2023, those banks that failed were characterised by lower degrees of diversification.
There’s another factor that also explains the downfall of LTCM, one that similarly has echoes in the banking sector. At the end of August, LTCM was bruised but it was far from bankrupt. It had working capital of around $4 billion including a largely unused credit facility of $900 million, of which only $2.1 billion was being used for financing positions.
But the fax Meriwether sent clients on September 2 triggered a run on the bank. “We had 100 investors at the time, and a couple of fax machines,” recalls Rosenfeld. “By the time we got to investor 50, I noticed that the top story on Bloomberg was us… All eyes were on us. We were like this big ship in a small harbour trying to turn; everyone was trying to get out of the way of us.”
While the August losses reflected a flight to quality as investors flocked to safe assets, the September losses reflected a flight away from LTCM. The price of a natural catastrophe bond the firm held, for example, fell by 20% on September 2, even though there had been no increase in the risk of natural disaster and the bond was due to mature six weeks later. As the firm was forced to divulge more information to counterparties over the course of September, the situation worsened. “The few things we had on that the market didn’t know about came back quickly,” Meriwether later told the New York Times. “It was the trades that the market knew we had on that caused us trouble.”
In addition, illiquid markets gave counterparties leeway in how to mark positions, and they used the opportunity to mark against LTCM to the widest extent possible so that they would be able to claim collateral to mitigate against a possible default (the flipside of the “two way mark-to-market”). The official inquiry into the failure noted that by mid-September, “LTCM’s repo and OTC [over-the-counter] derivatives counterparties were seeking as much collateral as possible through the daily margining process, in many cases by seeking to apply possible liquidation values to mark-to-market valuations.” And because different legs of convergence trades were held with different counterparties, there was very little netting. In index options, such collateral outflows led to around $1 billion of losses in September.
Nicholas Dunbar, who wrote the other bestselling book about LTCM, Inventing Money, quotes a trader at one of LTCM’s counterparties (emphasis added):
“When it became apparent they [LTCM] were having difficulties, we thought that if they are going to default, we’re going to be short a hell of a lot of volatility. So we’d rather be short at 40 [at an implied volatility of 40% per annum] than 30, right? So it was clearly in our interest to mark at as high a volatility as possible. That’s why everybody pushed the volatility against them, which contributed to their demise in the end.”
The episode is a lesson in endogenous risk. It’s a risk that differentiates securities markets from other domains governed by probability. “The hurricane is not more or less likely to hit because more hurricane insurance has been written,” mused one of LTCM’s partners afterwards. “In the financial markets this is not true. The more people write financial insurance, the more likely it is that a disaster will happen, because the people who know you have sold the insurance can make it happen. So you have to monitor what other people are doing.”
4. Why the Era of Historically Low Interest Rates Could Be Over – Nick Timiraos
At issue is what is known as the neutral rate of interest. It is the rate at which the demand and supply of savings is in equilibrium, leading to stable economic growth and inflation.
First described by Swedish economist Knut Wicksell a century ago, neutral can’t be directly observed. Instead, economists and policy makers infer it from the behavior of the economy. If borrowing and spending are strong and inflation pressure rising, neutral must be above the current interest rate. If they are weak and inflation is receding, neutral must be lower.
The debate over where neutral sits hasn’t been important until now. Since early 2022, soaring inflation sent the Federal Reserve racing to get interest rates well above neutral.
With inflation now falling but activity still firm, estimates of the neutral rate could take on greater importance in coming months. If neutral has gone up, that could call for higher short-term interest rates, or delay interest-rate cuts as inflation falls. It could also keep long-term bond yields, which determine rates on mortgages and corporate debt, higher for longer…
…Analysts see three broad reasons neutral might go higher than before 2020.
First, economic growth is now running well above Fed estimates of its long-run “potential” rate of around 2%, suggesting interest rates at their current level of 5.25% and 5.5% simply aren’t very restrictive.
“Conceptually, if the economy is running above potential at 5.25% interest rates, then that suggests to me that the neutral rate might be higher than we’ve thought,” said Richmond Fed President Tom Barkin. He said it is too soon to come to any firm conclusions.
That said, a model devised by the Richmond Fed, which before the pandemic closely tracked Williams’s model, put the real neutral rate at 2% in the first quarter.
Second, swelling government deficits and investment in clean energy could increase the demand for savings, pushing neutral higher. Joseph Davis, chief global economist at Vanguard, estimates the real neutral rate has risen to 1.5% because of higher public debt…
…Third, retirees in industrial economies who had been saving for retirement might now be spending those savings. Productivity-boosting investment opportunities such as artificial intelligence could push up the neutral rate.
And business investment depreciates faster nowadays and is thus less sensitive to borrowing costs, which would raise neutral. It is dominated by “computers and software, and much less office buildings, than it used to be,” Summers said during a lecture in May…
…Fed Chair Jerome Powell has in the past warned against setting policy based on unobservable estimates such as neutral, which he compared to navigating by the celestial stars.
Last December, he said the Fed would be careful about fine-tuning interest rates based on such estimates—for example, because falling inflation pushes real rates well above neutral. “I don’t see us as having a really clear and precise understanding of what the neutral rate is and what real rates are,” Powell said.
Some economists reconcile the debate by differentiating between short-run and longer-run neutral. Temporary factors such as higher savings buffers from the pandemic and reduced sensitivity to higher rates from households and businesses that locked in lower borrowing costs could demand higher rates today to slow the economy.
But as savings run out and debts have to be refinanced at higher rates in the coming years, activity could slow—consistent with a neutral rate lower than it is now.
5. Defining, Measuring, and Managing Technical Debt – Ciera Jaspan and Collin Green
We took an empirical approach to understand what engineers mean when they refer to technical debt. We started by interviewing subject matter experts at the company, focusing our discussions to generate options for two survey questions: one asked engineers about the underlying causes of the technical debt they encountered, and the other asked engineers what mitigations would be appropriate to fix this debt…
…This provided us with a collectively exhaustive and mutually exclusive list of 10 categories of technical debt:
- Migration is needed or in progress: This may be motivated by the need to scale, due to mandates, to reduce dependencies, or to avoid deprecated technology.
- Documentation on project and application programming interfaces (APIs): Information on how your project works is hard to find, missing or incomplete, or may include documentation on APIs or inherited code.
- Testing: Poor test quality or coverage, such as missing tests or poor test data, results in fragility, flaky tests, or lots of rollbacks.
- Code quality: Product architecture or code within a project was not well designed. It may have been rushed or a prototype/demo.
- Dead and/or abandoned code: Code/features/projects were replaced or superseded but not removed.
- Code degradation: The code base has degraded or not kept up with changing standards over time. The code may be in maintenance mode, in need of refactoring or updates.
- Team lacks necessary expertise: This may be due to staffing gaps and turnover or inherited orphaned code/projects.
- Dependencies: Dependencies are unstable, rapidly changing, or trigger rollbacks.
- Migration was poorly executed or abandoned: This may have resulted in maintaining two versions.
- Release process: The rollout and monitoring of production needs to be updated, migrated, or maintained.
We’ve continued to ask engineers (every quarter for the last four years) about which of these categories of technical debt have hindered their productivity in the previous quarter. Defying some expectations, engineers do not select all of them! (Fewer than 0.01% of engineers select all of the options.) In fact, about three quarters of engineers select three or fewer categories. It’s worth noting that our survey does not ask engineers “Which forms of technical debt did you encounter?” but only “Which forms of technical debt have hindered your productivity?” It’s well understood that all code has some technical debt; moreover, taking on technical debt prudently and deliberately can be a correct engineering choice.4 Engineers may run into more of these during the course of a quarter, but their productivity may not be substantially hindered in all cases.
The preceding categories of technical debt have been shown in the order of most to least frequently reported as a hindrance by Google engineers in our latest quarter. We don’t expect this ordering to generalize to other companies as the ordering probably says as much about the type of company and the tools and infrastructure available to engineers as it does the state of the code base. For example, Google engineers regularly cite migrations as a hindrance, but large-scale migrations are only attempted at all because of Google’s monolithic repository and dependency system;5 other companies may find that a large-scale migration is so impossible that it is not even attempted. A fresh start-up might have few problems with dead/abandoned code or code degradation but many hindrances due to immature testing and release processes. While we do expect there to be differences across companies in how much engineers are hindered by these categories, we believe the list itself is generalizable.
Our quarterly engineering survey enables us to measure the rate at which engineers encounter and are hindered by each type of technical debt, and this information has been particularly useful when we slice our data for particular product areas, code bases, or types of development. For example, we’ve found that engineers working on machine learning systems face different types of technical debt when compared to engineers who build and maintain back-end services. Slicing this data allows us to target technical debt interventions based on the toolchain that engineers are working in or to target specific areas of the company. Similarly, slicing the data along organizational lines allows directors to track their progress as they experiment with new initiatives to reduce technical debt.
However, we find quarterly surveys are limited in their statistical and persuasive power…
…Our goal was then to figure out if there are any metrics we can extract from the code or development process that would indicate technical debt was forming before it became a significant hindrance to developer productivity. We ran a small analysis to see if we could pull this off with some of the metrics we happened to have already…
…The results were disappointing, to say the least. No single metric predicted reports of technical debt from engineers; our linear regression models predicted less than 1% of the variance in survey responses. The random forest models fared better, but they had high precision (>80%) and low recall (10%–25%). That is, these models could identify parts of the code base where a focused intervention could reduce technical debt, but they were also going to miss many parts of the code base where engineers would identify significant issues.
It is quite possible that better technical debt indicator metrics do exist for some forms of technical debt. We only explored objective metrics for three types of technical debt, and we only sought to use existing metrics, rather than attempting to create new metrics that might better capture the underlying concepts from the survey.
However, it’s also possible that such metrics don’t exist for other types of technical debt because they are not about the present state of a system, but a relation between the system’s present state and some unimplemented ideal state. An engineer’s judgments about technical debt concern both the present state and the possible state. The possible states of the world are something that mathematical models cannot incorporate without the modeler’s direct intervention. For example, the fact that a project’s code base consists entirely of code written in Python 2 is not technical debt in a world where there is no loss of functionality compared to another language or version or outside pressure to migrate. However, in a world where Python 3 is a preferred or required alternative, that same corpus of Python 2 constitutes a needed migration. The present state of the world—from the perspective of a model—is identical in these two instances, but the possible world has changed. Humans consider the possible world in their judgments of technical debt. If a model were to incorporate explicit rules that capture aspects of the possible world (for example, if a model were designed to count every file in Python 2 as technical debt because the human modeler knows Python 3 is an alternative), then the change would be detectable to the model. If we could capture this judgment as it evolves, it could form the basis for better measurements of technical debt…
…While we haven’t been able to find leading indicators of technical debt thus far, we can continue to measure technical debt with our survey and help to identify teams that struggle with managing technical debt of different types. To that end, we also added the following questions to our engineering survey:
- To what extent has your team deliberately incurred technical debt in the past three months?
- How often do you feel that incurring technical debt was the right decision?
- How much did your team invest in reducing existing technical debt and maintaining your code?
- How well does your team’s process for managing technical debt work?
Combined with the survey items about the types of technical debt that are causing productivity hindrances, these questions enable the identification of teams that are struggling, reveal the type(s) of technical debt they are struggling with, and indicate whether they are incurring too much debt initially or whether they are not adequately paying down their existing debt. These are useful data, especially when teams can leverage them under guidance from experts on how to manage their technical debt. Fortunately, we have such experts at Google. Motivated in part by our early findings on technical debt, an interested community within Google formed a coalition to help engineers, managers, and leaders systematically manage and address technical debt within their teams through education, case studies, processes, artifacts, incentives, and tools. The coalition’s efforts have included the following:
- Creating a technical debt management framework to help teams establish good practices. The framework includes ways to inventory technical debt, assess the impact of technical debt management practices, define roles for individuals to advance practices, and adopt measurement strategies and tools.
- Creating a technical debt management maturity model and accompanying technical debt maturity assessment that evaluates and characterizes an organization’s technical debt management process and helps grow its capabilities by guiding it to a relevant set of well-established practices for leads, managers, and individual contributors. The model characterizes a team’s maturity at one of four levels (listed here from least to most mature):
- Teams with a reactive approach have no real processes for managing technical debt (even if they do occasionally make a focused effort to eliminate it, for example, through a “fixit”). Teams with a proactive approach deliberately identify and track technical debt and make decisions about its urgency and importance relative to other work.
- Teams with a strategic approach have a proactive approach to managing technical debt (as in the preceding level) but go further: designating specific champions to improve planning and decision making around technical debt and to identify and address root causes.
- Teams with a structural approach are strategic (as in the preceding level) and also take steps to optimize technical debt management locally—embedding technical debt considerations into the developer workflow—and standardize how it is handled across a larger organization.
- Organizing classroom instruction and self-guided courses to evangelize best practices and community forums to drive continual engagement and sharing of resources. This work also includes a technical talk series with live (and recorded) sessions from internal and external speakers.
- Tooling that supports the identification and management of technical debt (for example, indicators of poor test coverage, stale documentation, and deprecated dependencies). While these metrics may not be perfect indicators, they can allow teams who already believe they have a problem to track their progress toward fixing it.
Overall, our emphasis on technical debt reduction has resulted in a substantial drop in the percentage of engineers who report that their productivity is being extremely to moderately hindered by technical debt or overly complicated code in their project. The majority of Google engineers now feel they are only “slightly hindered” or “not at all hindered” by technical debt, according to our survey. This is a substantial change and, in fact, is the largest trend shift we have seen in five years of running the survey.
Disclaimer: None of the information or analysis presented is intended to form the basis for any offer or recommendation. We currently have a vested interest in Alphabet (parent of Google). Holdings are subject to change at any time.
