Global

Highlights It may seem self-evident that most governments are overly indebted, but both theory and evidence suggest otherwise. Higher debt today does not require higher taxes tomorrow if the growth rate of the economy exceeds the interest rate on government bonds. Not only is that currently the case, but it has been the norm for most of history. Unlike private firms or households, governments can choose the interest rate at which they borrow, provided that they issue debt in their own currencies. Ultimately, inflation is the only constraint to how large fiscal deficits can get. Today, most governments would welcome higher inflation. There are increasing signs China is abandoning its deleveraging campaign. Fiscal policy will remain highly accommodative in the U.S. and will turn somewhat more stimulative in Europe. Remain overweight global equities/underweight bonds. We do not have a strong regional equity preference at the moment, but expect to turn more bullish on EM versus DM by the middle of this year. Feature A Fiscal Non-Problem? Debt levels in advanced economies are higher today than they were on the eve of the Global Financial Crisis. Rising private debt accounts for some of this increase, but the lion’s share has occurred in government debt (Chart 1). Chart 1Global Debt Levels Have Risen, Especially In The Public Sector Not surprisingly, rising public debt levels have elicited plenty of consternation. While there has been a lively debate about how fast governments should tighten their belts, few have disputed the seemingly self-evident opinion that some degree of “fiscal consolidation” is warranted. Given this consensus view, one would think that the economic case for public debt levels being too high is airtight. It’s not. Far from it. Debt Sustainability, Quantified Start with the classic condition for debt sustainability, which specifies the primary fiscal balance (i.e., the overall balance excluding interest payments) necessary to maintain a constant debt-to-GDP ratio (See Box 1 for a derivation of this equation).   An increase in the economy’s growth rate (g), or a decrease in real interest rates (r), would allow the government to loosen the primary fiscal balance without causing the debt-to-GDP ratio to increase (Chart 2).1 If the government were to ease fiscal policy beyond that point, debt would rise in relation to GDP. But by how much? It is tempting to assume that the debt-to-GDP ratio would then begin to increase exponentially. However, that is only true if the interest rate is higher than the growth rate of the economy. If the opposite were true, the debt-to-GDP ratio would rise initially but then flatten out at a higher level.2 A Fiscal Free Lunch The last point is worth emphasizing. As long as the interest rate is below the economic growth rate, then any primary fiscal balance – even a permanent deficit of 20%, or even 30% of GDP – would be consistent with a stable long-term debt-to-GDP ratio. In such a setting, the government could just indefinitely rollover the existing stock of debt, while issuing enough new debt to cover interest payments. No additional taxes would be necessary. In fact, stabilizing the debt-to-GDP ratio becomes easier the higher it rises. Chart 3 shows this point analytically.    Ah, one might say: If the government issues a lot of debt, then interest rates would rise, and before we know it, we are back in a world where the borrowing rate is above the economy’s growth rate, at which point the debt dynamics go haywire. Now, that sounds like a sensible statement, but it is actually quite misleading. As long as a government is able to issue its own currency, it can always create money to pay for whatever it purchases. If people want to turn around and use that money to buy bonds, they are welcome to do so, but the government is under no obligation to pay them the interest rate that they want. If they do not wish to hold cash, they can always use the cash to buy goods and services or exchange it for foreign currency. As long as a government is able to issue its own currency, it can always create money to pay for whatever it purchases. Wouldn’t that cause inflation and currency devaluation? Yes, it might, and that’s the real constraint: What limits the ability of governments with printing presses to run large deficits is not the inability to finance them. Rather, it is the risk that their citizens will treat their currencies as hot potatoes, rushing to exchange them for goods and services out of fear that rising prices will erode the purchasing power of their cash holdings. When Is Saving Desirable? The reason governments pay interest on bonds is because they want people to save more. However, more savings is not necessarily a good thing. This is obviously the case when an economy is depressed, but it may even be true when an economy is at full employment. Just like someone can work so much that they have no time left over for leisure, or buy a house so big that they spend all their time maintaining it, it is possible for an economy to save too much, leading to an excess of capital accumulation. Under such circumstances, steady-state consumption will be permanently depressed because so much of the economy’s resources are going towards replenishing the depreciation of the economy’s capital stock.  Economists have a name for this condition: “dynamic inefficiency.” What determines whether an economy is dynamically inefficient? As it turns out, the answer is the same as the one that determines whether debt ratios are on an explosive path or not: The difference between the interest rate and the economy’s growth rate. Economies where interest rates are below the growth rate will tend to suffer from excess savings. In that case, government deficits, to the extent that they soak up national savings, may increase national welfare.   r < g Has Been The Norm Today, the U.S. 10-year Treasury yield stands at 2.69%, compared to the OECD’s projection of nominal GDP growth of 3.8% over the next decade. The gap between projected growth and bond yields is even greater in other major economies (Chart 4). Granted, equilibrium real rates are likely to rise over the next few years as spare capacity is absorbed. Structural factors might also push up real rates over time. Most notably, the retirement of baby boomers could significantly curb income growth, leading to a decline in national savings. Chart 5 shows that the ratio of workers-to-consumers globally is in the process of peaking after a three-decade long ascent. Economic growth could also fall if cognitive abilities continue to deteriorate, a worrying trend we discussed in a recent Special Report.3 Chart 5The Global Worker-To-Consumer Ratio Has Peaked It may take a while before real rates rise above GDP growth. Still, it may take a while before real rates rise above GDP growth. As Olivier Blanchard, the former chief economist at the IMF, noted in his Presidential Address to the American Economics Association earlier this year, periods in U.S. history where GDP growth exceeds interest rates have been the rule rather than the exception (Chart 6).4 The same has been true for most other economies.5 Chart 6GDP Growth Above Interest Rates: Historically, The Rule, Not The Exception What’s Next For Fiscal Policy? Austerity fatigue has set in. In the U.S., fiscally conservative Republicans, if they ever really existed, are a dying breed. Trump’s big budget deficits and his “I love debt” mantra are the waves of the future. For their part, the Democrats are shifting to the left, with the “Green New Deal” proposal being the latest manifestation. The case for fiscal stimulus is stronger in the euro area than for the United States. The European Commission expects the euro area to see a positive fiscal thrust of 0.40% of GDP this year, up from a thrust of 0.05% of GDP last year (Chart 7). This should help support growth. Chart 7The Euro Area Will Benefit From A Modest Amount Of Fiscal Easing This Year Additional fiscal easing would be feasible. This is clearly true in Germany, but even in Italy, the cyclically-adjusted government primary surplus is larger than what is necessary to stabilize the debt ratio.6 Unfortunately, the situation in southern Europe is greatly complicated by the ECB’s inability to act as an unconditional lender of last resort to individual sovereign borrowers. When a government cannot print its own currency, its debt markets can be subject to multiple equilibria. Under such circumstances, a vicious spiral can develop where rising bond yields lead investors to assign a higher default risk, thus leading to even higher yields (Chart 8).   Mario Draghi’s now-famous “whatever it takes” pledge has gone a long way towards reassuring bond investors. Nevertheless, given the political constraints the ECB faces, it is doubtful that Italy or other indebted economies in the euro area will be able to pursue large-scale stimulus. Instead, the ECB will keep interest rates at exceptionally low levels. A new round of TLTROs is also looking increasingly likely, which should protect against a rise in bank funding costs and a potential credit crunch. Our European team believes that a TLTRO extension would be particularly helpful to Italian banks.  Even in Italy, the cyclically-adjusted government primary surplus is larger than what is necessary to stabilize the debt ratio. Despite having one of the highest sovereign debt ratios in the world, Japan faces no pressing need to tighten fiscal policy. Instead of raising the sales tax this October, the government should be cutting it. A loosening of fiscal policy would actually improve debt sustainability if, as is likely, a larger budget deficit leads to somewhat higher inflation (and thus, lower real borrowing rates) and, at least temporarily, faster GDP growth. We expect the Abe government to counteract at least part of the sales tax increase with new fiscal measures, and ultimately to abandon plans for further fiscal tightening over the next few years. In the EM space, Brazil, Turkey, and South Africa are among a handful of economies with vulnerable fiscal positions. They all have borrowing rates that exceed the growth rate of the economy, cyclically-adjusted primary budget deficits, and above-average levels of sovereign debt (Chart 9).   In contrast, China stands out as having the biggest positive gap between projected GDP growth and sovereign borrowing rates of any major economy. The problem is that the main borrowers have been state-owned companies and local governments, neither of which are backstopped by the state. Not officially, anyway. Unofficially, the government has been extremely reluctant to allow large-scale defaults anywhere in the economy. Despite all the rhetoric about market-based reforms, they are unlikely to start now. Historically, the Chinese government has allowed credit growth to reaccelerate whenever it has fallen towards nominal GDP growth. As we recently argued in a report entitled “China’s Savings Problem,” China needs more debt to sustain aggregate demand.7 Historically, the government has allowed credit growth to reaccelerate whenever it has fallen towards nominal GDP growth (Chart 10). The stronger-than-expected jump in credit origination in January suggests that we are approaching such an inflection point. Chart 10Historically, China Has Scaled Back On Deleveraging When Credit Growth Has Fallen Close To Nominal GDP Growth Investment Conclusions The consensus economic view is that deflation is a much harder problem to overcome than inflation. When dealing with inflation, all you have to do is raise interest rates and eventually the economy will cool down. With deflation, however, a central bank could very quickly find itself up against the zero lower bound constraint on interest rates, unable to ease policy any further via conventional means. While this standard argument is correct, it takes a very monetary policy-centric view of macroeconomic policy. When interest rates are low, fiscal policy becomes very potent. Indeed, the whole notion that deflation is a bigger problem than inflation is rather peculiar. Just as it is easier to consume resources than to produce them, it should be easier to get people to spend than to save. People like to spend. And even if they didn’t, governments could go out and buy goods and services directly. Looking out, our bet is that policymakers will increasingly lean towards the ever-more fiscal stimulus. If structural trends end up causing the so-called neutral rate of interest to rise – the rate of interest that is necessary to avoid overheating – policymakers will have no choice but to eventually raise rates and tighten fiscal policy (Box 2). However, they will only do so begrudgingly. The result, at least temporarily, will be higher inflation. Fixed-income investors should maintain below benchmark duration exposure over both a cyclical and structural horizon. Reflationary policies that increase nominal GDP growth will help support equities, at least over the next 12 months. Chart 11 shows that corporate earnings tend to accelerate whenever nominal GDP growth rises. We upgraded global equities to overweight following the December FOMC meeting selloff. While our enthusiasm for stocks has waned with the year-to-date rally, we are sticking with our bullish bias. Chart 11Earnings And Nominal GDP Growth Tend To Move In Lock-Step A reacceleration in Chinese credit growth will put a bottom under both Chinese and global growth by the middle of this year. As a countercyclical currency, the dollar will likely come under pressure in the second half of this year. Until then, we expect the greenback to be flat-to-modestly stronger. The combination of faster global growth and a weaker dollar later this year will be manna from heaven for emerging markets. We closed our put on the EEM ETF for a gain of 104% on Jan 3rd, and are now outright long EM equities. I do not have a strong view on the relative performance of EM versus DM at the moment, but expect to shift EM equities to overweight by this summer.8 Peter Berezin, Chief Global Strategist Global Investment Strategy peterb@bcaresearch.com   Box 1 The Arithmetic Of Debt Sustainability   Box 2 Debt Sustainability And Full Employment: The Role Of Fiscal And Monetary Policy Policymakers should strive to stabilize the ratio of debt-to-GDP over the long haul, while also ensuring that the economy stays near full employment. The accompanying chart shows the tradeoffs involved. The DD schedule depicts the combination of the primary fiscal balance and the gap between the borrowing rate and GDP growth (r minus g) that is consistent with a stable debt-to-GDP ratio. In line with the debt sustainability equation derived in Box 1, the slope of the DD schedule is simply equal to the debt/GDP ratio. Any point below the DD schedule is one where the debt-to-GDP ratio is rising, while any point above is one where the ratio is falling. The EE schedule depicts the combination of the primary fiscal balance and r - g that keeps the economy at full employment. The schedule is downward-sloping because an increase in the primary fiscal balance implies a tightening of fiscal policy, and hence requires an offsetting decline in interest rates. Any point above the EE schedule is one where the economy is operating at less than full employment. Any point below the EE schedule is one where the economy is operating beyond full employment and hence overheating. Suppose there is a structural shift in the economy that causes the neutral rate of interest – the rate of interest consistent with full employment and stable inflation – to increase. In that case, the EE schedule would shift to the right: For any level of the fiscal primary balance, the economy would need a higher interest rate to avoid overheating. The arrows show three possible “transition paths” to a new equilibrium. Scenario #1 is one where policymakers raise rates quickly but are slow to tighten fiscal policy. This results in a higher debt-to-GDP ratio. Scenario #2 is one where policymakers tighten fiscal policy quickly but are slow to raise rates. This results in a lower debt-to-GDP ratio. Scenario #3 is one where the government drags its feet in both raising rates and tightening fiscal policy. As the economy overheats, real rates actually decline, sending the arrow initially to the left. This effectively allows policymakers to inflate away the debt, leading to a lower debt-to-GDP ratio. Note: In Scenario #2, and especially in Scenario #3, the DD line will become flatter (not shown on the chart to avoid clutter). Consequently, the final equilibrium will be one where real rates are somewhat higher, but the primary fiscal balance is somewhat lower, than in Scenario #1.   Footnotes 1          One can equally define the interest rate and GDP growth rate in nominal terms (see Box 1 for details).  2       Japan is a good example of this point. The primary budget deficit averaged 5% of GDP between 1993 and 2010, a period when government net debt rose from 20% of GDP to 142% of GDP. Since then, Japan’s primary deficit has averaged 5.1% of GDP, but net debt has risen to only 156% of GDP (and has been largely stable for the past two years). 3      Please see Global Investment Strategy Special Report, “The Most Important Trend In The World Has Reversed And Nobody Knows Why,” dated February 1, 2019. 4      Olivier Blanchard, “Public Debt And Low Interest Rates,” Peterson Institute for International Economics and MIT American Economic Association (AEA) Presidential Address, (January 2019). 5      Paolo Mauro, Rafael Romeu, Ariel Binder, and Asad Zaman, “A Modern History Of Fiscal Prudence And Profligacy,” IMF Working Paper, (January 2013). 6      The Italian 10-year bond yield is 2.83% while nominal GDP growth is 2.64%. Multiplying the difference by net debt of 118% of GDP results in a required primary surplus of .22% of GDP that is necessary to stabilize the debt-to-GDP ratio. This is lower than the IMF’s 2018 estimate of cyclically-adjusted government primary surplus of 2.14%. 7      Please see Global Investment Strategy Weekly Report, “China’s Savings Problem,” dated January 25, 2019. 8      Please note that my colleague, Arthur Budaghyan, BCA’s Chief EM strategist, remains bearish on both EM and DM equities and expects EM to underperform DM over the coming months. Strategy & Market Trends MacroQuant Model And Current Subjective Scores Tactical Trades Strategic Recommendations Closed Trades        

Special Report

When we first set out to create ETS, we leveraged the vast amount of research on stock market anomalies and started with factors that had a proven track-record. We also wanted to minimize the risk of data mining, and as a result, we decided to keep things simple. We stuck with established metrics, and avoided venturing down the long, windy path of custom metrics and adjustments: the more parameters that are included, the easier it is to find spurious results. We also opted for simple, linear mappings – lower is better, or higher is better. When deviating from this approach, it is easy to lose sight of the big picture and spend time optimizing minute details that will not have a large impact on the overall performance of the model. Fast-forward to 2019. We have a stable model that has performed well out of sample, and we are now comfortable taking a second pass over our factors. We can now analyze each model input in detail to see if we can make any improvements. The first factor under the microscope is the Altman Z-Score. This metric was first developed by Edward I. Altman in 1968 to estimate the probability of bankruptcy over a two-year horizon. It uses various items from a firm’s income statements and balance sheet and combines them using a set of weights to yield the so-called Altman Z-Score.1 Since this metric relies on estimated coefficients for a specific set of data, and given that these coefficients date back to the 60s, we wanted to investigate the possibility of using a more modern model of distress in place of the original Altman-Z score.  Our research process led us to review the academic literature, replicate the most promising results internally, and finally, to implement our own variation of the Altman-Z score given our unique dataset.   A Proxy For Bankruptcy In order to assess the ability of a distress model to predict insolvency, it is necessary to have a historical dataset of failure events. If not, it is impossible to measure the success of a given model. Failure can be defined in several ways, the most obvious of which is a Chapter 7 or Chapter 11 bankruptcy filing, at least in the U.S. To ensure that we captured the maximum possible failure events for our global set of stocks, we avoided the Chapter 7/11 definition and developed a simple market-cap based proxy for bankruptcy. We define the failure event of a firm as the point in time where the market-cap drops to 1% of its historical maximum. We also enforce that this event must occur after the maximum to prevent false failures at the beginning of a firm’s trading history. We found that this definition signals failure at the appropriate time for a selection of large-cap firms known to have experienced distress; a notable example is Enron Corp. (Chart 1). In general, we deem this a reasonable proxy for bankruptcy since the price of shares will drop substantially upon news of insolvency. Ultimately, even though it may, in some cases, flag firms that are technically not bankrupt, for our purposes, we are happy to avoid buying firms that plunge below 1% of their previous valuation peak. Academia To The Rescue? A review of the modern finance literature suggested that the Altman Z-Score, while methodologically sound, still had room for improvement as a distress measure. We anticipated that using a more refined distress model, we could amplify the effect of the existing Altman-Z Factor in ETS. In light of this, our first attempt at replacing the Altman-Z score involved running a logistic regression (logit) model presented in Campbell et al. (2010)2 on the ETS universe of stocks and constructing a new factor based on the output from this model. Following the authors’ methodology, we constructed a set of eight variables (three accounting ratios, and five market-based variables) and computed the probability of failure on our entire historical dataset. The probabilities were then percentile-ranked to construct a score from 0-100%, where firms with a score of 0% correspond to those with the highest probability of failure. This ranking system is consistent with the other ETS factors where lower scores are bad and higher scores are good. Although the model by Campbell et al. worked well for predicting subsequent failure (as judged by our bankruptcy proxy measure), the metric fell short performance-wise when tested within our universe of stocks. In brief, our decile-spread metric3 yielded a negative number, indicating poor separation of returns between deciles. Based on the above, we sought to construct our own distress model using similar methods, but applied to the universe of valid ETS firms. There are several advantages to this approach: First, it prevents errors in the construction of the explanatory variables. For instance, when replicating models found in academic literature, we cannot be sure that we are constructing our variables in the exact same way as reported, and hence we cannot have complete confidence in the accuracy of the computed probabilities of failure. Second, it frees us from the shackles of a complicated model, and we can seek to reduce our own model down to something more computationally tractable, and avoid overfitting the data.  Finally, it allows us to base our model on global stock data, as opposed to the U.S.-centric CRSP/COMPUSTAT dataset used by many academic institutions. Bringing It “In-House” The construction of our own distress model began with an analysis of the variables defined in Campbell et al. First, we wanted to determine which variables showed the strongest deviation between success and failure groups. From the calculation of summary statistics and histograms, the variables that showed the most prominent separations between success and failure groups were Net Income to Market Total Assets (NIMTA), Total Debt to Market Total Assets (TDMTA), monthly excess return (EXRET), and the 3-month standard deviation of daily returns (SIGMA) (Chart 2). Detailed descriptions of the variables are provided in the Appendix. We computed NIMTA, TDMTA, EXRET, and SIGMA at monthly frequencies with the goal of constructing a logit model that would predict failure 12 months into the future. The logit model implies that at time t, the probability of firm i failing in the next 12 months is given by:   where y represents the sequence of success (0) and failure (1) events for each firm at each point in time, x represents the four explanatory variables, and α, β are the parameters to fit. The parameters were fit on a rolling basis to avoid look-ahead bias when constructing a trading strategy based on the outcome of the model. Therefore, at each year-end, we used the trailing success/failure data up to that point to estimate parameters. The fitting yielded intuitive results, suggesting that higher TDMTA, SIGMA and lower NIMTA, EXRET indicate a higher probability of failure (see parameter list in Appendix Table 1). Another positive characteristic of these parameters is that they remain relatively stable throughout time, alluding to the robustness of the model. From visual inspection of the parameters, we see that EXRET has the strongest effect on failure probability, followed by SIGMA, TDMTA, and NIMTA. It is not surprising that EXRET has the strongest effect since it is based on a 12-month trailing weighted average. Therefore, we can expect this variable to capture the prolonged period of descent that firms typically undergo before reaching a bankruptcy event. Indeed, firms in our failure group experienced, on average, a failure event 5 years after reaching peak market cap. As for the remaining variables, it is natural to expect increased volatility and subpar financial statements to signal imminent failure. Model accuracy was tested by looking at its predictions on an individual and aggregate basis. Reassuringly, we found individual examples of large firms with known failure events showing the expected behavior (Chart 3). However, the more significant result is that on average, the model predicts a high probability of failure for the group of true failure firms (N=3213) relative to a random sample of true success firms (Chart 4). Armed with a successful distress model, we then translated the predictions into an ETS factor, giving it a score from 0-100%. Using a relative change condition in the parameters as a guide,4 we chose the factor inception date to be year-end 2004. Since our stock data begins in 1995, this allows for a sufficient “burn-in” time for the distress model to stabilize. The latter is important given the spike in U.S. Chapter 11 filings starting in 2001, which is also captured by our bankruptcy proxy (Chart 5).  With the factor constructed, we replaced the existing Altman-Z factor in the full model, keeping the same initial weight.  As expected, this replacement led to an improvement in the overall model according to our standard performance metrics, which are described in a previous report.5 Specifically, we observed a 6% improvement in the decile ordering metric (DECORD) and a small (<1%) reduction in mean absolute difference in BCA Score (MAD), leading to a 7% increase in our overall risk-reward ratio (DECORD / MAD). The new factor, simply named Distress on the platform, also performs well when examined in isolation. To test this, we ran the equivalent of an equal-weighted, daily-rebalanced backtest on ten deciles separated by the Distress score. Calculating the Sharpe ratio on the decile returns, we observe an increasing trend in the performance of the factor as a function of decile (Chart 6). Therefore, with this factor we obtain the desired separation in performance between deciles on a risk-adjusted basis.  What About Machine Learning? Disclaimer: This section involves a small digression on machine learning One statistical challenge with our success/failure classification problem is that the number of “success” events (i.e. no indication of bankruptcy) greatly outnumbers the failure events. Indeed, with our bankruptcy proxy, approximately 1% of the 3.5 million firm-month observations in our dataset are marked as failures. Despite the small failure rate, classic logistic regression can still provide good results; however, we wished to explore the effectiveness of machine learning techniques in tackling this type of classification problem. The use of machine learning (ML) was motivated by an analogous problem outside the world of stocks, namely the detection of credit card fraud. One can imagine that out of the massive number of credit card transactions registered daily, only a small fraction is actually fraudulent. Now, suppose that for a given period we obtained transaction data that includes characteristics such as dollar amount, time of day, and whether or not the transaction was normal or fraudulent. We could then use this prior data to train an ML model to recognize fraudulent transactions when presented with new, out-of-sample data. A specific class of ML models known as “autoencoders” have been shown to be effective in handling this task (See Box 1 for more details). Given that the credit-card problem is directly analogous to our distress problem, we set up a simple autoencoder to test whether it could detect firm failure with a higher degree of accuracy than the logit model. As inputs to the model, we used the same eight variables from Campbell et al. Perhaps surprisingly, we found that the best version of our autoencoder model did not surpass the accuracy of the logit model, at least in terms of maximizing precision and recall simultaneously.6 Furthermore, the predictions of the autoencoder were less effective at separating winners from losers after being transformed into a factor score between 0-100%. This, of course, does not mean that ML techniques are hopeless. It simply means that, for this particular set of parameters and model structure, the autoencoder was not able to provide additional predictive value over the simpler, more computationally efficient logit model.     Box 1 Autoencoders The rise of machine learning has brought with it a slew of jargon that seems intimidating at first, but can actually be understood without extensive training in mathematics or computer science. The term “autoencoder” may very well fall into this category. To get right into it, an autoencoder is a type of neural network tasked with learning a compressed representation of the original input or “training” data. The flow of data through an autoencoder is shown below (Box Diagram 1). The goal of the autoencoder is to optimize the weights connecting nodes in the network so that the difference between the input and output is minimized. However, due to the action of the “reduction” or “encoding” phase of the autoencoder, the output, i.e., the “reconstructed” or “decoded” data, will always be a compressed approximation of the original input. The latter is important for classification problems because it means that post-training, any data presented to the network that does not resemble the original training data will likely be poorly reconstructed, and flagged as anomalous.  Going back to our problem of identifying distressed firms, let’s step through how we use an autoencoder to distinguish between “success” and “failure” groups. Separate the data (i.e., the observations for all accounting and market-based variables) into success and failure groups according to the bankruptcy proxy measure. Train the autoencoder to reconstruct the success group. Each variable of interest corresponds to a node in the “input” layer of the network. Feed the autoencoder the failure group data and record the error in reconstructing this data. Set an error threshold (a number) for determining when a firm should be marked a failure vs. success. Generate predictions based on this threshold. Analyze the accuracy of these predictions using metrics such as precision and recall. Future Work A key priority of the ETS team is to use modern techniques in computing to uncover new sources of alpha for our clients. With the recent explosion of advancements in data science and machine learning, we are excited about leveraging the knowledge from these fields to improve the ETS model. We believe that this report provides a glimpse into the style of factor analysis we expect to conduct in the near future. As for the distress model developed here, we are content with the performance of the simple logistic regression for now, but we will continue to examine whether modern classification techniques can help improve accuracy.  One method in particular that piques our interest is so-called “Gradient Boosting”, which has made big waves in the data science community for its use in winning competitions on Kaggle.com.7 Overall, we will continue to monitor the model’s ability to exploit stock-market anomalies and adjust accordingly, whether that means adjustments to existing factors, or the development of new ones. In addition, we will strive to keep ETS an integral part of your asset management workflow via new features on the web platform. As always, we hope you find the recent additions valuable, and please do not hesitate to reach out with any comments or questions.   Spencer Moran, Senior Analyst Equity Trading Strategy spencerm@bcaresearch.com   Appendix Variable Definitions Here we provide the definitions of the variables used in the distress model. The denominator in NIMTA and TDMTA represents “Market Total Assets” and is given by Market Cap. plus Total Debt.  The formulas are as follows: EWM is a 12-month exponential weighted moving average used to incorporate more history into the variable but provide greater weight to more recent observations. STD is an annualized, trailing three-month standard deviation. The symbols R, Rm, and r represent monthly returns, monthly market returns, and daily returns, respectively. Parameters Here we provide the parameters (aka a and b’s) obtained from the rolling fit of the distress model. Recall that parameters are estimated at the end of each year, and then these parameters are used to compute predictions for the following year to avoid look-ahead bias. Footnotes 1      The original Z-score formula was as follows: Z =   1.2X1 + 1.4X2 + 3.3X3 + 0.6X4 + 1.0X5, where: X1 : working capital / total assets X2 : retained earnings / total assets X3 : earnings before interest and taxes / total assets  X4 : market value of equity / book value of total liabilities X5 : sales / total assets 2  Campbell, John Y., Jens Dietrich Hilscher, and Jan Szilagyi. 2011. Predicting financial distress and the performance of distressed stocks. Journal of Investment Management 9(2): 14-34. 3   The decile-spread is a metric we use to evaluate the performance of a trading system. Essentially, it looks at the sum of the differences between the annual compounded growth rates of each score decile. Ideally, the deciles should be in order, and the upper deciles should outperform the lower deciles. This metric captures both these aspects, and summarizes them in a single number. 4       The factor inception date was chosen to be the date when the relative change in the magnitude of the parameter set (treated as a vector) fell below 1%. 5      Please see Equity Trading Strategy, “ETS Goes Global,” dated October 31, 2016, available at ets.bcaresearch.com 6      Precision is defined as the ratio of actual failures to all failures predicted by the model. Meanwhile, recall is the ratio of actual failures to the sum of actual failures plus false successes predicted by the model. 7      Kaggle is a popular online community (recently acquired by Google) where members compete to solve open problems in data science and machine learning.

Highlights Global Growth: Early leading indicators (credit impulses, our global LEI diffusion index) are signaling that the worst of the global economic downturn should soon end. Okun’s Law: In the developed economies, the observed relationships between economic growth and changes in unemployment suggest that the current pullback in global growth will not be severe enough to create slack in labor markets and reduce inflation pressures. Global Bond Allocation: Within dedicated global government bond portfolios, stay underweight the U.S. and Canada, neutral core Europe, and overweight the U.K., Japan and Australia. Remain tactically overweight global credit versus government bonds, at least until mid-year, with policymakers likely to stay cautiously dovish until global uncertainties recede. Feature Is This Risk Rally Too Good To Last? The mood of financial markets has improved significantly over the past few weeks, led by the dovish shift from central bankers that has revived investor risk appetite. Some positive headlines on U.S.-China trade negotiations have also generated hope over prospects for a deal, further fueling the bullish sentiment. The global economic picture remains muddled, though. Non-U.S. growth continues to languish, while the actual near-term state of the U.S. economy is proving difficult to determine given the data issues surrounding the 35-day U.S. government shutdown. Given lingering uncertainties, both political and economic, policymakers do not want to rock the boat by saying anything that might be interpreted as hawkish. With monetary policy no longer a near-term headwind, there is a window for continued outperformance of global risk assets in the next few months. That means higher global equity prices and stable-to-tighter global corporate credit spreads. Yet the seeds for the next wave of market turbulence may already be sewn. There are signs that the global growth downturn may soon end. Credit impulses are starting to pick up in several major economies, while our diffusion index of global leading economic indicators – itself a longer leading indicator – has clearly bottomed (Chart of the Week). The epicenter of global economic weakness, China, continues to deploy monetary and fiscal stimulus measures aimed at stabilizing growth. Meanwhile, the U.S. economy still appears to be in good shape, underpinned by solid consumer fundamentals. Chart of the WeekSunnier Days Ahead? A combination of easier financial conditions and faster economic growth will eventually prove to be incompatible with stable monetary policy, especially with surprisingly firm inflation in the major developed economies. Central bankers will respond by moving away from their current dovish bias, led by the U.S. Federal Reserve. With government bond markets now discounting both stable monetary policy and too-low inflation expectations, the path for global bond yields is eventually higher. While headline inflation rates are cooling in response to the lagged impact of weaker oil prices, the pullback has been far more muted so far compared to similar sharp oil-driven moves in the past (Chart 2). This is because domestically-driven inflation rates for services and wages are much sturdier today in many countries. If BCA’s bullish oil view for 2019 comes to fruition, then the current decline in headline/goods inflation rates may prove to be very short-lived and with little pass-through into core/services inflation. Chart 2Sticky Global Inflation, Despite Lower Oil Prices This dynamic is not the same in every country, however. When looking at the individual trends of goods inflation and services/wage inflation in the major developed economies, the largest gaps between the two exist in the U.S. and Canada (Chart 3). There, wage growth is accelerating and services inflation rates remain sturdy, despite sharp drops in goods inflation. Chart 3Domestic Inflation Pressures Most Acute In The U.S. & Canada Our recommended government bond allocation at the country level reflects these underlying inflation trends. We are more bearish on bond markets with the most intense domestic inflation pressures – and where future interest rate hikes are most likely – and vice versa. We remain underweight the U.S. and Canada, where wage growth and services inflation are both above the inflation targets of the Fed and Bank of Canada, and where market-based measures of inflation expectations like CPI swap rates have already bottomed (Chart 4). We remain neutral on core Europe (Germany, France) where wage growth has perked up, core/services inflation remains closer to 1% than the 2% target of the ECB, and inflation expectations continue to drift lower. Finally, we remain overweight the U.K., Japan and Australia, all of which have an underlying inflation picture that is muted enough to keep policymakers on hold for at least the next 6-9 months. Chart 4Favor Bond Markets Where Domestic Inflation Pressures Are Weakest Bottom Line: Early leading indicators (credit impulses, our global LEI diffusion index) are signaling that the worst of the global economic downturn should soon end. Central bankers will remain cautious and dovish in the near-term, however, implying that the current outperformance of global equity and credit markets has more room to run – but also setting up the next upleg for bond yields later this year. Okun’s Law Revisited Central bankers remain wedded to the idea that there is an “exploitable” relationship between unemployment and inflation, a.k.a. the Phillips Curve. A logical extension is that unless policymakers can credibly forecast a reduction in labor demand that pushes unemployment rates beyond levels associated with full employment, inflation will not be expected to decline. Policymakers will have a difficult time staying dovish without believing that inflation pressures are diminishing. One way to measure the relationship between economic growth and changes in economic slack is by using a concept that you may remember from an old macroeconomics class – Okun’s Law. More an empirically observable rule of thumb than any rule based in actual economic theory, Okun’s Law simply measures how much unemployment rates change relative to swings in real GDP growth. Past estimations for the U.S. economy have shown that the long-run coefficient in the Okun’s Law regression is around 2, which means that a 2% fall in real GDP growth should be associated with a 1% increase in the unemployment rate (and vice versa). That coefficient is not the same over shorter time horizons, though, as the unemployment/GDP growth relationship can be impacted by other cyclical factors like changes in hours worked or labor productivity. Charts 5 and 6 show annual real GDP growth (the percentage change over four quarters) versus the change in the unemployment rate over twelve months for the major developed economies (the U.S., U.K., euro area, Japan, Canada, Australia, New Zealand and Sweden) dating back to 1980. There is a reasonably strong relationship between the two series in the charts, although the “fit” does vary from country to country. Chart 5The Okun’s Law Relationship … Chart 6… Still Holds For Most Countries That can be seen in the individual country scatterplots shown in Charts 7 to 14, which plot each quarterly data point of the change in unemployment and real GDP growth. The darker dots represent the period from 1980-2010, while the lighter dots are the post-2010 era. The actual estimated regression, and its R-squared, are also shown in the charts (the equation can be defined as “the estimated change in the unemployment rate for a given pace of real GDP growth”). For most countries shown, the R-squareds are reasonably good (between 0.55 and 0.70) for a single-factor model like this. The coefficients on the change in real GDP are all between -0.35 and -0.45, which means that a fall in real GDP growth of 3.5 to 4.5 percentage points is consistent with a rise in the unemployment rate of 1 percentage point. The lone country where the Okun’s Law relationship has a relatively poor historical fit is in Japan, which is due to the lack of GDP variability relative to swings in the unemployment rate, especially over the past decade. We can use these estimates of the Okun’s Law coefficient to conduct a “back of the envelope” thought experiment that answers the following question that relates to the current economic and financial market backdrop: how much of a decline in GDP growth is necessary to raise unemployment rates back to full-employment (NAIRU) levels? As we have consistently noted in recent Weekly Reports, global central bankers can only turn so dovish, even after the severe market turbulence seen at the end of last year and with elevated political uncertainty in many locations. Why? Because unemployment rates remain below levels that are consistent with stable inflation. Without a meaningful weakening of labor markets that pushes unemployment rates back above “full employment” levels, policymakers will not be able to lower their inflation forecasts and signal a need for easier monetary policy. In Table 1, we present the estimated Okun’s Law regressions from 1980, along with the real GDP growth rate that falls out of those equations if we assume the employment gaps are closed.1 We also show the consensus 2019 real GDP growth forecasts taken from Bloomberg, as well as the expected change in central bank policy rates over the next year taken from our Central Bank Discounters. The conclusion from the Table is that it would take significant declines in real GDP growth to raise unemployment rates enough for policymakers to become less worried about inflation pressures. Table 12019 Consensus Growth Forecasts Are Well Above Levels That Would Eliminate The Unemployment Gap In the U.K., where the unemployment rate is furthest below the OECD’s estimate of the full-employment NAIRU rate, a whopping -3.3 percentage point cut to real GDP growth is needed to raise unemployment back to 5.6%. The required GDP fall is lower in the U.S., with only a -1.6 percentage point decline in real GDP growth need to push the unemployment rate back to the OECD NAIRU estimate of 4.3%. Falls in real GDP growth of between -1.5 and 2.0 percentage points are necessary in most of the other countries to close the “unemployment gap”, except for Japan. Given the weak estimated Okun’s Law relationship in Japan, we are reluctant to put much weight on the results of this thought experiment for Japan. Those “required” declines in real GDP growth are nowhere close to the 2019 consensus Bloomberg forecasts for each country. This is even true in the U.S., where the consensus expects real GDP growth to decline by -0.9 percentage points in 2019. Unsurprisingly, markets are discounting very little change in monetary policy over the next year according to our Central Bank Discounters, with modest odds of a rate cut now discounted in Australia (-19bps), New Zealand (-11bps) and the U.S. (-8bps) and a full 25bp hike now priced in Sweden. Summing it all up, our simple Okun’s Law thought experiment shows that it would take a significantly larger decline in global growth than the consensus, or BCA, expects for central banks to shift even more dovishly in the direction of interest rate cuts. This puts a cyclical floor underneath global bond yields, given that relatively stable policy rates are now discounted. Bottom Line: The observed relationships between economic growth and changes in unemployment suggest that the current pullback in global growth will not be severe enough to create slack in labor markets and an easing of inflation pressures in the developed economies.   Robert Robis, CFA, Senior Vice President Global Fixed Income Strategy rrobis@bcaresearch.com   Footnotes 1 Given the declining productivity trend seen in all countries over the past 20 years, we have made a downward adjustment to those Okun’s Law estimated coefficients. In other words, we do not think that it will take the same magnitude of GDP loss to generate the same increase in unemployment when labor productivity is low. Recommendations The GFIS Recommended Portfolio Vs. The Custom Benchmark Index Duration Regional Allocation Spread Product Tactical Trades Yields & Returns Global Bond Yields Historical Returns