Inflation

If we look at global growth as an aircraft, the plane is experiencing failing engines and will lose more altitude in the coming months. Yet, neither Chinese authorities, nor the Fed or the ECB will be quick to come to the rescue as global growth downshifts. These dynamics herald a stronger US dollar and lower EM risk asset prices.

The broader rally that started in June is premised on a Goldilocks narrative that will prove to be a fairy tale. Either by stubborn inflation. Or, by higher unemployment that shows that the war on inflation is far from costless. Or, by both. We discuss the implications for stocks and bonds. And we reveal our new top long dollar cross.

Highlights The simple relationship between the unemployment rate and inflation, the traditional Phillips Curve, is not especially strong. But this perspective is archaic and ignores the lessons learned by central bankers during the 1960s and 1970s. Investors need to make three adjustments to this simple unemployment rate / core inflation relationship to arrive at the “modern-day” Phillips Curve. Expectations for inflation need to be included in some form, the unemployment rate should be smoothed and measured relative to some estimate of equilibrium, and variables that capture trade effects should be incorporated. Doing so vastly improves the ability to explain inflation over the past six decades. Our estimation of the “modern-day” Phillips Curve highlights a great irony of the recent debate over the Curve: the mystery is why inflation was so high during the first half of the last expansion, not why inflation was so low during the second half. The monetary policy implications of our estimation of the “modern-day” Phillips Curve are straightforward: barring an exogenous shift lower in long-term inflation expectations or some disinflationary trade event, it is unlikely that the Fed will see a sustainable return to its 2% target without a rise in the unemployment rate above NAIRU. Such an increase in the unemployment rate has always been associated with a US recession in the post-WWII era. The Fed’s implication that it may begin to cut interest rates if it gains confidence that inflation is moving back toward its 2% target is either a reference to small technical adjustments within the context of still-tight monetary policy, or it is a policy mistake that could cause the outcome that the Fed is desperate to avoid – structurally high inflation caused by too easy monetary policy and unanchored long-term inflation expectations. We are likely to recommend significantly boosting exposure to inflation protection if we see concrete signs that the Fed is likely to cut interest rates prematurely. If that scenario materializes and inflation reaccelerates, investors will have few options at their disposable to earn positive absolute returns. International investors should favor the US dollar in a premature rate cut scenario, whereas US-based investors should significantly increase their holdings of very short-term Treasurys. Feature Since the global financial crisis, it has become a stylized economic fact that the Phillips Curve has flattened significantly. Policymakers spent much of the last economic cycle either downplaying the Phillips Curve or declaring it dead. While Fed Chair Jerome Powell noted last week at Jackson Hole that “there is evidence that inflation has become more responsive to labor market tightness than was the case in recent decades”, he also signaled a degree of agnosticism about the Phillips Curve in stating that “these changing dynamics may or may not persist.”1 When thinking about the Phillips Curve as a simple relationship between today’s unemployment rate and today’s inflation rate (especially headline inflation), we agree that the link is not especially strong. But this is hardly a new phenomenon. And this perspective entirely ignores the lessons that were learned by central bankers during the 1960s and 1970s that came to re-characterize the relationship between inflation and unemployment as the expectations-augmented (or “modern-day”) Phillips Curve.2 In addition to incorporating the impact of inflation expectations, most expressions of the modern-day Phillips Curve also factor-in trade effects, given the very significant rise in global trade as a share of GDP that has occurred over the past 50 years. In this report, we mostly reject the idea that the link between unemployment and inflation is much weaker than it has been in the past and present a modern-day Phillips Curve model that does a very good job at explaining inflation since the 1960s. We review and explain the few cases in which the model’s accuracy was below-average, and note that the model’s failure to predict inflation fully from 2009-2019 was mostly concentrated in the first half of that period. In fact, the model highlights a great irony of the debate over the Phillips Curve: the mystery is why inflation was so high during the first half of the last expansion, not why inflation was so low during the second half. We conclude that the Fed’s implication that it may begin to cut interest rates if it gains more confidence that inflation is moving back toward its 2% target is either a reference to small technical adjustments within the context of still-tight monetary policy, or it is a policy mistake that could cause the outcome that the Fed is desperate to avoid – structurally high inflation caused by too easy monetary policy and unanchored long-term inflation expectations. We are likely to recommend significantly boosting exposure to inflation protection if we see concrete signs that the Fed is likely to cut interest rates before the US unemployment rate begins to rise. International investors should favor the US dollar in a premature rate cut scenario, whereas US-based investors should significantly increase their holdings of very short-term Treasurys. Modeling The Modern-Day Phillips Curve Charts II-1 and II-2 effectively summarize the basis for challenging the relevance of the traditional Phillips Curve. Chart II-1 shows a scatterplot of core PCE inflation as a function of the unemployment rate from 1960 to 2019, and shows an inconsistent relationship (at best). Chart II-2 removes the high inflation / high unemployment rate era of the 1970s and 1980s, and shows that the relationship is somewhat improved. However, there are still many observations that do not follow the expected negative relationship between unemployment and inflation, and both charts show that the dots associated with the 2000-2019 period look almost like a flat line. From our perspective, investors need to make three adjustments to this simple unemployment rate / core inflation relationship to arrive at the “modern-day” Phillips Curve. Expectations for inflation need to be included in some form, the unemployment rate should be smoothed and measured relative to some estimate of equilibrium, and variables that capture trade effects should be incorporated. In statistical terms, this means that inflation must be modeled using a multivariate process, rather than by comparing inflation to a single variable as is done in Charts II-1 and II-2. Chart II-1The Traditional Phillips Curve… September 2023 September 2023 Chart II-2…Does Not Work Very Well September 2023 September 2023 Chart II-3 presents the results of these adjustments, and shows that they vastly improve the ability to explain inflation over the past six decades. The technical details of our estimation of the modern-day Phillips Curve are described in Box II-1, but the important points are as follows: We model the deviation of core inflation from our measure of inflation expectations, instead of including inflation expectations as a predictor of inflation. We use the 10-year average of headline inflation as our measure of inflation expectations. We use the CBO’s estimate of NAIRU as our equilibrium measure of the unemployment rate. We include two variables to measure the effects of global trade on inflation: import price inflation, and the five-year change in US import intensity. These variables are effective at capturing trade effects, especially during the 1990s when core inflation fell below the Fed’s target. The smoothed unemployment rate gap is the single most important variable in the model, followed closely by import prices. The relationship shown in Chart II-3 is not perfect, underscoring that the model highlights a tendency for inflation rather than a hard forecast. But deviations in actual inflation from the model’s predicted value are generally explainable through the lens of the three variables. Panel 2 of Chart II-3 presents the residual of the model, which only shows sustained divergences between actual inflation and the model’s value during the late-1960s, the mid-to-late-1980s, 2011-2015, and the pandemic era. Notably, the model performed well in the latter half of the last economic expansion, which is precisely when policymakers and many investors derided the Phillips Curve as being largely or mostly irrelevant. We discuss each of these episodes below, and explain why we think actual inflation was different than the model forecasted during these periods. Chart II-3The Modern-Day Phillips Curve, However, Works Quite Well The Modern-Day Phillips Curve, However, Works Quite Well The Modern-Day Phillips Curve, However, Works Quite Well BOX II-1 The Technical Details Of Our Modern-Day Phillips Curve Estimation The dotted line in panel 1 of Chart II-3 illustrates our estimate of the modern-day US Phillips Curve. The series is derived from an ordinary least-squares regression, with the difference between the year-over-year growth rate in the core PCE deflator and the 10-year moving average in headline inflation as the dependent variable (i.e., core inflation relative to a simple inflation expectations proxy). We include the following three independent variables in the model: the three-year average of the CBO’s estimate of NAIRU minus the actual unemployment rate, the year-over-year growth rate in non-oil import prices (backfilled with the overall import price deflator prior to 1967), and the five-year annualized rate of change of the ratio of real imports to real GDP (inverted). In the model, the unemployment rate gap represents traditional Phillips Curve effects, and import prices and the long-term change in import intensity reflects trade effects. The model’s adjusted r-squared value is 55%, and the t-statistics of the three variables are 10.8, 10.5, and 8.1, respectively. As the model estimates the difference in core inflation from our simple inflation expectations proxy, panel 1 of Chart II-3 shows the fitted value of the model added to the 10-year moving average of headline inflation. Concerning non-oil import prices, we do not include relative import prices because Chart II B-1 shows that they have historically led core inflation. So, when including the simple year-over-year growth rate in non-oil import prices, it is unlikely that we are just picking up a spurious correlation (where some common factor driving both import prices and US core inflation). Chart II B-1Import Prices Reliably Lead US Core Inflation September 2023 September 2023 The Late-1960s Chart II-3 shows that actual inflation was generally lower than the model’s forecast in the first half of the 1960s, but the magnitude of the difference was smaller than it was in the second half of that decade and the model’s value led actual inflation. By contrast, inflation during the late-1960s was significantly higher than what our estimation of the modern-day Phillips Curve would have implied, which does require an explanation. The best explanation available for the surprisingly high inflation seen during the late-1960s is that the modern-day Phillips Curve was “kinked” during this period. By this, we mean that there was an asymmetric relationship between the unemployment rate gap and inflation that depended on whether the gap was positive or negative. It is also possible that the late-1960s represented a “catch-up” phase for inflation, given that actual inflation was below our model’s forecast during the first half of that decade. My colleague Peter Berezin, BCA’s Chief Global Strategist and Director of Research, has argued that the Phillips Curve is kinked in general, not just during the second half of the 1960s. Chart II-4 and Chart II-5 show that the relationship between inflation and the unemployment rate clearly inflected in the 1960s once the unemployment rate fell below 4% and that it has followed a somewhat similar path since the onset of the pandemic. Chart II-4The Unemployment Rate / Inflation Relationship Was Kinked In The 1960s… September 2023 September 2023 Chart II-5…And May Also Be Today September 2023 September 2023 In the view of the Bank Credit Analyst service, we agree that the relationship between inflation and unemployment does probably change when the unemployment rate is below NAIRU. So, in that regard, we agree that some of the modern-day Phillips Curve reflects a permanently kinked relationship between inflation and unemployment. However, we are not entirely convinced that this explains the entire post-pandemic inflation story, as we will detail below. The Mid-To-Late-1980s Chart II-3 shows that while our model correctly predicted the pickup in inflation that occurred in the second half of the 1980s, actual inflation was meaningfully and persistently below what our model suggests should have occurred during the mid-to-late-1980s. This episode is the easiest to explain and relates to our proxy for long term inflation expectations. As discussed in Box II-1, applying the modern-day Phillips Curve framework is analytically challenging because good measures of long-term inflation expectations do not exist prior to the 1980s. Even the measures that exist are not especially robust in their early years, as they consisted of surveyed responses that were either narrow in breadth or inconsistent in frequency. Long-term inflation expectations are crucially important to the modern-day Phillips Curve, as they represent the “baseline” for inflation that one would expect to prevail if the labor market and any other important driving factors of inflation were at neutral or average levels. As such, the use of proxies is necessary to model inflation properly. In the case of the mid-1980s the story is simply that our proxy for inflation expectations was “slower” than actual inflation expectations. Chart II-6Our Model Overestimated Inflation In The 1980s Because Actual Long-Term Inflation Expectations Fell Sharply Our Model Overestimated Inflation In The 1980s Because Actual Long-Term Inflation Expectations Fell Sharply Our Model Overestimated Inflation In The 1980s Because Actual Long-Term Inflation Expectations Fell Sharply Chart II-6 makes this point by showing our inflation expectations proxy alongside that of the median five-to-ten-year inflation expectation series from the University of Michigan’s Survey of Consumers. While long-term consumer inflation expectations are not especially well-anchored in terms of their level, changes in the series are highly informative. The chart highlights that surveyed long-term consumer inflation expectations fell very sharply during the 1980s, more quickly than a ten-year moving average of inflation. This explains why the model value shown in Chart II-3 was persistently above actual inflation during that period: the true baseline for inflation was lower than our expectations proxy suggested. The Modern-Day Phillips Curve During The Last Economic Expansion The first half of the last economic expansion is the most significant challenge to our estimation of the modern-day Phillips Curve, and does support the idea that the Curve was “flatter” during that period. Based on how severely the unemployment rate rose, inflation should have been much weaker during this period given the historical relationship between the two variables and after controlling for trade effects. In Charts II-1 and II-2, this period shows up as the group of observations at the far bottom right of the charts, which do indeed contribute to the flatness of the relationship observed post-2000. There are three potential explanations for why inflation did not fall as much as it should have in the first half of the last expansion. The first possibility is that NAIRU rose significantly during this period, meaning that the unemployment rate gap was not actually as large as current estimates of NAIRU would imply. This is a possibility, but we are strongly inclined to view the 2009-2015 period as a time of significantly elevated and long-lasting cyclical unemployment rather than an increase in structural unemployment. Thus, the very elevated unemployment rate should have severely weighed on wages and thus lowered core inflation to much lower levels than those that prevailed from 2011 to 2015. The second potential explanation is that the relationship between inflation expectations and actual inflation strengthened during this period due to the Fed’s strong record of maintaining low and stable inflation (thus preventing a disinflationary spiral). In addition, the fact that the Fed actively communicated to the public during the early recovery years that a large part of its objective was to prevent deflation may have helped support prices. For example, in a CBS interview following the Fed’s November 2010 decision to engage in a second round of quantitative easing (“QE2”), then-Chair Ben Bernanke prominently tied the decision to the fact that “inflation is very, very low.” When asked whether additional rounds of easing might be required, Bernanke responded that it was “certainly possible” and again cited inflation as a core consideration. The third potential explanation relates to how firms responded to the subpar growth recovery that occurred in the first half of the last economic expansion. The theoretical impact of changes in the unemployment rate and inflation occurs through wages. Chart II-7 highlights that nominal wage growth was indeed extremely weak from 2010 to 2014. What is strange is that this did not fully affect core consumer prices. We think that part of this occurred because firms were reluctant to cut prices in the face of very weak top-line growth. Chart II-8 highlights that S&P 500 sales per share growth was quite low during this period. We suspect that firms compensated for slow revenue growth by holding the growth in selling prices constant as wage growth fell (thus boosting margins, see panel 2 of Chart II-8). Chart II-7A Very High Unemployment Rate Did Severely Depress Wage Growth Last Cycle A Very High Unemployment Rate Did Severely Depress Wage Growth Last Cycle A Very High Unemployment Rate Did Severely Depress Wage Growth Last Cycle Chart II-8Firms Resisted Cutting Prices Last Cycle So They Could Make Up For Weak Top Line Growth Firms Resisted Cutting Prices Last Cycle So They Could Make Up For Weak Top Line Growth Firms Resisted Cutting Prices Last Cycle So They Could Make Up For Weak Top Line Growth     A broader question about this period is why the recovery continued at all given its subpar nature, i.e., why did firms continue to hire in the face of such powerful headwinds to growth. We return to Fed credibility to answer this question: the Fed maintained a hyper-accommodative monetary policy during this period, provided additional policy responses to growth shocks like the euro area sovereign debt crisis, and actively persuaded investors and economic agents that it would do everything in its power to ensure a continued expansion of the economy. The Fed’s jawboning ended up succeeding, but it resulted in an erosion of the Fed’s forecasting credibility as well as a period of painfully weak real wage growth. As noted above, one very important point for investors to understand is that the re-emergence of the flat Phillips Curve narrative happened especially in the latter half of the last expansion, as part of an attempt to explain why inflation was not higher despite a low unemployment rate. But as Chart II-3 shows, aside from modestly lower inflation than would be expected in 2019 (see Box II-2), the mystery is actually why inflation was not weaker in the first half of the expansion, and that the inflation rate observed in the second half of the expansion was almost entirely normal. BOX II-2 Was The Behavior Of Core Inflation in 2019 Proof Of A Flat Phillips Curve? Our estimate of the modern-day US Phillips Curve suggests that US core inflation should have been above 2% and accelerating at the end of 2019 / beginning of 2020. Instead, actual core inflation decelerated from about 2% in late-2018 to roughly 1.5% by the end of 2019, in the face of a sub-4% unemployment rate. Was this evidence of a weak or irrelevant relationship between inflation and unemployment? In our view, the answer is no. The reason is that our estimation of the modern-day Phillips Curve highlighted the importance of trade effects as a driver of inflation, and 2018 and 2019 featured a major trade shock: the Trump administration’s trade policy. As a result of the imposition of tariffs against China and several developed market economies, US industrial production contracted, the new orders component of the US manufacturing PMI fell below 50, and global import volume collapsed, all of which are very rare outside of the context of a recession. Businesses were hyper-aware of the risks to the business cycle from the imposition of tariffs, and trade uncertainty exploded (Chart II B-2). Price-setting was likely impacted by this event, even though it did not raise the unemployment rate. Chart II B-2Inflation Slowed In 2019 Because Of A Massive Trade Shock Inflation Slowed In 2019 Because Of A Massive Trade Shock Inflation Slowed In 2019 Because Of A Massive Trade Shock Chart II-9Massive And Unexpected US Shale Production Caused A Huge Shock To Oil Prices In 2014 Massive And Unexpected US Shale Production Caused A Huge Shock To Oil Prices In 2014 Massive And Unexpected US Shale Production Caused A Huge Shock To Oil Prices In 2014 How is it then that our model was forecasting at-target inflation in the second-half of the last economic expansion despite a very low unemployment rate? Does that not fly completely in the face of the idea of the Phillips Curve and justify arguments from investors and policymakers that the Phillips Curve is no longer relevant? The experience of the second half of the last economic expansion does explain why the traditional Phillips Curve lacks the ability to explain inflation effective today. But the modern-day Phillips Curve is alive and well, and the reason that inflation was not above the Fed's target in the few years prior to the pandemic is because inflation expectations fell to levels that were not target-consistent. Charts II-9 to II-11 illustrate this point. While inflation did not fall relative to expectations in the first half of the last expansion as much as the unemployment rate would have implied, the long-lasting weakness in demand left expectations vulnerable to exogenous shocks. In 2014, such a shock occurred: oil prices collapsed almost exactly at the point that US tight oil production crossed the four-million-barrels-per-day mark (Chart II-9), a level of output that many experts had previously believed would not be attainable (or would roughly mark the peak in production). We view this event as a truly exogenous shock to prices, given that research and development of shale technology had been ongoing since the late 1970s and only happened finally to gain traction around 2010. Chart II-10 shows that the 2014 oil price collapse caused a clear break lower in the measure of inflation expectations that we used in our modern-day Phillips Curve estimate, to the lowest value recorded since the early 1960s. This break also occurred in market-based expectations of inflation, such as long-dated CPI swap rates and TIPS breakeven inflation rates, and surveys of consumer inflation expectations (Chart II-11). This decline in inflation expectations meant that the unemployment rate needed to be below NAIRU just for the Fed to hit its 2% target (absent any upwards shock to prices). It also meant that the meaningful acceleration of inflation from 2016 to 2018 should be viewed as inflation “outperformance” because its long-term trend had been lowered by the earlier downward shift in expectations. Chart II-10The 2014 Oil Shock Caused A Huge Decline In Our Model's Inflation Expectations Proxy... September 2023 September 2023 Chart II-11...As Well As Any Available Measure Of Inflation Expectations ...As Well As Any Available Measure Of Inflation Expectations ...As Well As Any Available Measure Of Inflation Expectations Chart II-11 also highlights that the pandemic fully normalized long-term inflation expectations, especially five-to-ten-year median household inflation expectations. This underscores that investors should expect future inflation to be higher than what occurred in the second half of the last economic expansion at any given unemployment rate, which has significant policy implications for the Fed. The Monetary Policy Implications Of The Modern-Day Phillips Curve When presenting Chart II-5 above, we noted that the relationship between inflation and unemployment has followed a pattern that is similar to what occurred in the late-1960s, giving credence to the idea that the relationship between the two is kinked. We agree that the tightening in the labor market over the past two probably explains some of the residual between actual inflation and our model’s estimate, but the supply-side of the economy has also played a very significant role: Chart II-12Supply-Side Effects Played An Important Role In The Surge In Inflation During The Pandemic Supply-Side Effects Played An Important Role In The Surge In Inflation During The Pandemic Supply-Side Effects Played An Important Role In The Surge In Inflation During The Pandemic Chart II-12 highlights that the pandemic caused an astronomical rise in global shipping costs and snarled the global supply chain. Russia’s invasion of Ukraine also likely delayed some of the normalization in the global supply chain, and significantly raised energy costs in Europe. Rolling lockdowns and other pandemic-related effects (such as producer/consumer order co-ordination failure in some industries and a surge in demand in others) caused a period of scarce supply for critical production components like semiconductors. This still has some residual effects today in terms of the production and inventory level of motor vehicles in the US. Social shifts during the pandemic, such as the prevalence of work-from-home policies, clearly impacted the housing market in the early phase of the pandemic and has worsened a housing shortage problem that existed even before the pandemic. Work-from-home effects also contributed to the semiconductor shortage, given the surge in tech demand that occurred as workers increasingly shifted to home offices (worker computing equipment, server demand to support virtual meetings, etc). Given the uniqueness of the COVID-19 pandemic and its economic consequences, no realistic economic model would be capable of truly capturing these effects, beyond the inclusion of “dummy” variables (that simply represent the average increase in inflation above what the modern-day Phillips Curve would suggest). As such, we are left with some combination of supply-side factors and a kinked unemployment rate / inflation relationship to explain the rise in inflation above what our model would suggest. We lean more toward the former rather than the latter. Turning to the implications of our model for Fed policy, we think it is important for investors to understand that the more the current overshoot in inflation is explainable by a kinked unemployment rate / inflation relationship, the lower the odds are of a soft-landing economic outcome materializing. That is because the unemployment rate is still below most estimates of NAIRU, underscoring that inflation is likely to remain above the Fed’s target until a recession takes place. And even if all of the current overshoot in inflation is due to residual pandemic-related effects that will fully dissipate, Chart II-13 shows what will happen to our model’s estimate for inflation if the unemployment rate stays at its current level and we assume no change in trade effects (positive or negative) over the coming 18 months. The chart shows that inflation would be forecast to rise sustainably above the Fed’s target, to a high of 3-3.5% by the middle of next year. Chart II-13The Fed Is Unlikely To See A Sustainable Return To Its 2% Target Without A Higher Unemployment Rate September 2023 September 2023 Chart II-14If The Unemployment Rate Stays Where It Is, Inflation Will Likely Move Higher If The Unemployment Rate Stays Where It Is, Inflation Will Likely Move Higher If The Unemployment Rate Stays Where It Is, Inflation Will Likely Move Higher Chart II-14 illustrates the basis for the forecast shown in Chart II-13: if the unemployment rate stays at its current level, the three-year moving average of the unemployment rate gap will rise to the highest level since the late-1990s, without the strong disinflationary trade effects that were present at that time. The only other period that saw a similar labor market was the late-1960s, when inflation eventually accelerated significantly (as shown in Chart II-4). The monetary policy implications of Charts II-13 and II-14 are straightforward: barring an exogenous shift lower in long-term inflation expectations or some disinflationary trade event, it is unlikely that the Fed will see a sustainable return to its 2% target without a rise in the unemployment rate above NAIRU. Even based on the Fed’s long-run unemployment rate estimate (which is lower than the CBO’s), that would imply a rise in the unemployment rate above 4%. It is not clear whether an unemployment rate of 4.5% – which is what the Fed currently forecasts for the end of 2024 – will be enough to return inflation back to the Fed’s 2% target. What is clear is that the outcome would result in roughly 1.7 million Americans losing their jobs, and would represent a rise in the unemployment rate that has always been associated with a US recession in the post-WWII era (Chart II-15). Chart II-15Since WWII, A Meaningful Rise In The Unemployment Rate Has Always Coincided With A Recession Since WWII, A Meaningful Rise In The Unemployment Rate Has Always Coincided With A Recession Since WWII, A Meaningful Rise In The Unemployment Rate Has Always Coincided With A Recession Investment Conclusions From the perspective of the modern-day Phillips Curve, the Fed’s implication that it may begin to cut interest rates if it gains confidence that inflation is moving back toward its 2% target is either a reference to small technical adjustments within the context of still-tight monetary policy, or it is a policy mistake that could cause the outcome that the Fed is desperate to avoid – structurally high inflation caused by too easy monetary policy and unanchored long-term inflation expectations. In the latter case (what we would refer to as the “premature rate cut scenario”), it is possible that the initial reaction of financial markets will be positive / pro-risk, with both stocks and bonds rallying. However, if the estimate shown in Chart II-13 is correct, it would ultimately create the need for a much more significant rise in interest rates and the unemployment rate, and thus a much more severe recession to return inflation back to the Fed’s target. In other words, it would represent a repeat of the January to October 2022 experience for financial markets, but likely without the excess savings cushion that has helped prevent a recession over the past 18 months. Chart II-16 highlights that there would be few places for investors to hide in such a scenario. The chart shows the average monthly nominal return from a large set of financial assets, and underscores that very few assets earned a positive return during this period. Those that did mostly reflected rising energy prices, underscoring that the performance of those assets may not fare as well in the face of a second inflationary wave as their returns last year were boosted by Russia’s invasion of Ukraine. Chart II-16If The Fed Causes A Repeat Of The 2022 Experience, There Will Be Few Places For Investors To Hide. Favor The US Dollar And Cash. September 2023 September 2023 The bottom line for investors is that we are likely to recommend significantly boosting exposure to inflation protection if we see concrete signs that the Fed is likely to cut interest rates before the US unemployment rate begins to rise. If that scenario materializes and inflation reaccelerates, investors will have few options at their disposable to earn positive absolute returns. International investors should favor the US dollar in a premature rate cut scenario, whereas US-based investors should significantly increase their holdings of very short-term Treasurys, as significantly positive real interest rates would likely be needed to re-anchor long-term inflation expectations. Long-duration positions would eventually be warranted once monetary policy inevitably tips the US economy into a serious recession, but the initial financial market response would very likely see equities and long-maturity bonds selling off, as they did in the first ten months of 2022. Jonathan LaBerge, CFA Vice President The Bank Credit Analyst   Footnotes 1 Inflation: Progress and the Path Ahead, Jerome Powell, August 25th, 2023. 2 Please see The Bank Credit Analyst “The Modern-Day Phillips Curve, Future Inflation, And What To Do About It,” dated December 18, 2020, available at bca.bcaresearch.com

In Section I, we respond to the ongoing challenge to our view that the US economy is on a recessionary path. The available evidence overwhelmingly supports the notion that US monetary policy is tight, which argues against the “no landing” economic scenario. It also underscores that the recessionary clock is indeed ticking unless the monetary policy stance eases soon. The “soft landing” narrative remains improbable and may have been unduly boosted by artificially low inflation readings over the summer. Until concrete signs of the meaningful rate cuts emerge, we will continue to recommend that investors maintain defensive portfolio positions. In Section II, we review the “modern-day” Phillips Curve, and explain why it is unlikely that the Fed will see a sustainable return to its 2% target without a rise in the unemployment rate above NAIRU.

Stocks should continue to rally in the near term, but investors should prepare to turn more defensive towards the end of the year in advance of a recession in 2024.

The Fed and ECB talked a good game as they redoubled their commitments to returning core inflation to 2% p.a. at Jackson Hole. However, their outmoded inflation-fighting playbooks do not address supply tightness in commodity and energy markets, which keeps inflation risk elevated. The proposed expansion of the BRICS states seeks to capitalize on these trends, and supports efforts to weaken the centrality of the USD in global trade. We remain long commodity exposure via ETFs to retain exposure to energy and metals producers and refiners.

The stock market’s pre-eminent growth sector is not US tech, it is French luxuries. No other sector can compare with French luxuries’ massive and sustained pricing power. The risk for French luxuries is not a China slowdown, the risk is that the structural increase in super-wealth comes to an end. If anything though, the coming disruption from generative AI will boost super-wealth. Ironically therefore, the best investment play on generative AI might be French luxuries.

In this report, we assess the best opportunities in inflation-linked bonds in the major developed economies, based on trends in growth, inflation and the stance of monetary policies in each country. We conclude that the environment is turning more challenging for European inflation-linked bond performance versus nominal government bonds, while the opposite is true in Japan. In the US, US TIPS breakevens have likely peaked, particularly at the short end.

Commentators often use notions like debt deflation, balance sheet recession, and liquidity trap interchangeably. Yet, these are different concepts. This report develops a framework and provides a diagnosis of China’s economic malaise. A follow-up report will deal with what kind of treatment is needed for a recovery. As a trade, we recommend shorting the EM equity index.

Numerous divergences have opened up between global risk assets and global business cycle variables. These gaps are unsustainable, and odds are that the recoupling will occur to the downside with risk assets selling off.