”May you live in interesting times.” 1
By now, we are sure most of you have read about the latest development in large language models (LLMs), which are artificial intelligence systems designed to process and generate natural language text. While several versions are under development, OpenAI has received a great deal of attention of late with the release of ChatGPT-4. We thought it would be timely to try out GPT-4 and incorporate a few queries and responses into this letter. While GPT-4’s knowledge base is limited to data through September 2021, the system is impressive, providing almost immediate responses to questions. Nevertheless, we lightly edited some of the responses we deemed overly wordy or repetitive.
On to the markets, which continue to be dominated by the interdependent forces of inflation, interest rates, and economic growth (or contraction as in a recession). Inflation has continued to fall from its 9% peak in June 2022 with the most recent monthly reading at 5%. That is still too high relative to the Fed’s long-term target of 2%. The market is expecting just one more rate hike of .25% in May before the Fed pauses and begins cutting rates later in the year. Given the lag between monetary policy and its desired effect, most observers also predict an economic slowdown or mild recession, short-term pain which Chairman Powell is willing to accept to avoid the longer-term corrosive effects of inflation.
In 2022, those higher interest rates reduced the value of both bonds and stocks, hitting growth stocks particularly hard. In the first quarter, the pain shifted to the banking sector. Much has been written about the Silicon Valley Bank failure, so we won’t rehash the saga here except to point out that many other regional banks face similar challenges (as evidenced by the news of First Republic being taken over by the FDIC and sold to J.P. Morgan): vulnerable deposits and unrecognized losses on both investments and loans. In the first quarter of 2023 alone, $600 billion in deposits exited U.S. commercial banks in search of greater safety and/or higher interest rates.
Q: What is the impact of a large reduction in commercial bank deposits?
GPT-4: A large reduction in bank deposits can have significant impacts on the bank and the broader economy. Here are some potential consequences:
The most acute consequences, liquidity issues and bank failures, were largely limited by the Fed’s rescue actions although many (including notable investors Paul Singer and Ken Griffin) fear the “cure” may prove temporary and only heighten long-term risks in the form of moral hazard.
GPT-4: Moral hazard is a term used in economics and finance to describe a situation where one party takes risks because they know they will not bear the full cost of those risks. … in the context of banking and finance, moral hazard might occur when banks or financial institutions take on excessive risk because they know that if things go wrong, they may be bailed out by the government or the central bank. This can lead to a destabilization of the financial system and ultimately harm the broader economy.
We have already seen many of the other consequences GPT-4 listed emerge. First quarter 2023 earnings announcements by regional banks generally reflected deposit outflows, lower net interest margins, and tightening lending standards, although conditions are still far from a “credit crunch”. The full economic impact of tighter lending is yet to be seen although it could be significant as regional banks are estimated to account for 50% of all commercial and industrial lending and 40-70% (depending on the source) of commercial real estate loans. These developments make the Fed’s tightrope act far more precarious.
Speaking of commercial real estate (CRE), this is developing into a looming threat to both real estate equity investors and lenders as longer-term trends accelerated by COVID impact the value and use cases of many properties. This is how GPT-4 describes the potential impacts of COVID on CRE (again based on data through September 2021):
Sounds pretty on point, doesn’t it? Office is currently showing the most weakness with high profile owners (Brookfield, Blackstone, and Columbia Property Trust to name a few) defaulting on large mortgages. A more widespread real estate correction could be triggered by the expiration of tenant leases and/or loans (an estimated 10% of CRE loans will mature in the coming year), potentially adding further stress to bank balance sheets and securitized markets.
Another potential jolt is the debt ceiling which the Treasury has avoided breaching through “extraordinary measures” since January 2023. Unless Congress can agree to raise the limit, the Treasury will run out of flexibility in the next few months, possibly as soon as June, given that tax receipts came in lower than expected. Regardless of your position on the issue, the time for compromise is becoming very short even while many warn of dire consequences. GPT-4 identifies potential consequences as a default on debt obligations, government shutdown, economic downturn, international repercussions (given that the U.S. dollar is the world’s reserve currency), and loss of investor confidence. As all of these imply a higher cost of capital, the impact on markets would likely be very negative at least in the short-term.
The rates market (Treasurys) reflects these myriad concerns as the yield on the 10 year Treasury note declined from 3.96% at the end of February to 3.57% as of this writing. On the short end of the curve, interest rates typically reflect the fed funds rate, although there is unusual divergence between one-month and three-month Treasury bills as investors opt for the shorter-term securities to avoid getting caught up in the debt ceiling drama. According to the Wall Street Journal, “Surging demand has driven one-month T bill prices higher, sending the yield down to 3.313% from 4.675% at the end of March. Bills maturing in three months yield 5.105%—a record incentive for lending to the government for a couple months more.” The 3-month/10-year Treasury yield curve inversion (shorter-term interest rates higher than longer-term rates) hit its steepest magnitude since 1981. Historically, this has been a reliable recession indicator.
On the other hand, equity markets, credit spreads, and the VIX (volatility index) all appear to be sending more sanguine signals. What accounts for the disconnect? In the case of the equity markets, stock leadership is the narrowest since the 1990s according to J.P. Morgan. Although the S&P 500 Index gained 7.5% for the quarter, looking below the surface reveals that the entirety of the return can be explained by 28 stocks and just six stocks accounted for 53% of the return. In keeping with our AI theme, these six stocks are all LLM innovation companies which gained 45% as a group in the wake of ChatGPT’s popularity. We note that these stocks were probably due for a bounce after being down significantly last year, likely benefited from a lower discount rate, and may be viewed in the context of flight to quality due to relatively high certainty of cash flows.
In high yield land, the index gained 3.7% for the quarter, a result which can be attributed to optimism for a soft landing (i.e., the Fed is able to traverse the high wire without falling). Despite concerns about an economic slowdown, defaults remain muted as many borrowers took advantage of the long period of accommodative credit markets to term out debt in structures marked by low rates, long maturities, and lax covenants. As a result, the “wall of maturities” seems manageable until the 2025-2028 timeframe, delaying the potential reckoning in the credit markets.
Lest the litany of challenges facing the economy and markets is discouraging, we should point out that cycles are inevitable and ultimately healthy for the system, though not for every individual security. Cycles provide a mechanism to wring out market excesses and create opportunities for investors.
While still early, we see this pattern emerging across a number of areas. Even though a traditional distressed credit cycle does not appear imminent, our multi-strategy managers report greater price dispersion among securities, mispricing within capital structures, more attractive terms from those companies needing capital, and pockets of distress in non-traditional areas. Less accommodative conditions in general should provide a catalyst for fundamental differentiation among companies and benefit long/short strategies, an environment for which we have waited patiently as our managers have slogged through three years of a largely macro-driven market.
Within global equities, we note that large cap U.S. equities trade at valuations above 25-year averages. Despite the recent dominance of U.S. equities, high quality companies are domiciled all around the globe and are currently cheaper overseas. In fact, stock prices have been less problematic than currencies. As the Fed reaches the end of its hiking cycle and interest rate differentials narrow, we expect that headwind to abate. The two most recent quarters have manifested a reversal of dollar strength although we believe there is a lot of room to run. We continue to find what we believe are attractive opportunities in the private markets and encourage you to read the letters on Private Equity and Real Assets included in this quarter’s book for a fuller discussion.
One of the tired axioms of our profession is “You can’t eat relative performance.” We believe the phrase misses the point if you have at least an intermediate time horizon, as preserving capital in down markets leaves an investor with more capital to compound in recovering markets. Imagine the benchmark that you are measured against moving along through time … over the years, there are both good markets and bad. We pursue a defensive approach that tends to do better than the balanced benchmark when markets are soft and to do a little worse when markets are very strong. The past three years, which we refer to as the “COVID Period,” have been “interesting times” indeed. This period captures the COVID-induced panic in equity markets, the rapid recovery as governments intervened, the resultant surge in inflation, aggressive rate hiking which hammered stocks and bonds in 2022, and, most recently, a narrow market recovery. Such rapid adjustments have at times created treacherous conditions as hidden risks were uncovered and trading anomalies appeared. Protecting capital during those two major downdrafts was an significant determinant of investment performance in the “COVID Period”. This is what we seek to accomplish in the ebb and flow of markets. Our primary tools are diversification, hedging, active management, and patience.
Returning to LLM and Artificial Intelligence (AI), the topic has become divisive with some thrilled with the potential benefits to society while others fear the potential risks. A group of experts in the field and industry leaders wrote an open letter calling for a six-month pause in developing systems more powerful than ChatGPT-4. As is typical in the case of rapidly developing technologies, the regulatory regimes are behind the curve. It seems that AI is here to stay in a very big way, and we are unlikely to “put the genie back in the bottle.” For now, we’ll let ChatGPT-4 have the last word on whether humans should be concerned about AI.
One can only hope.
1 From GPT-4: “May you live in interesting times” is a well-known but apocryphal English expression that is often considered to be a curse, although it is also sometimes seen as a blessing. The phrase is believed to have originated in China, but there is no clear evidence of this. In any case, the phrase is often used to suggest that living in "interesting" or turbulent times can be both exciting and challenging, but also difficult and dangerous. It is important to note, however, that the phrase is not an actual Chinese proverb or curse, and it is not necessarily an accurate reflection of the Chinese view of history or the world.