May 2015 » White paper The “old normal” returns: Forecasting long-term U.S. growth potential Key takeaways Izzet Yildiz, Ph.D. Analyst Michael J. Atkin Portfolio Manager Jo Anne Ferullo, CFA Senior Investment Director • Contrary to widely held expectations of secular stagnation in U.S. economic growth, we expect the U.S. economy will reassert its strength over the long term. Fears of secular stagnation The pace and sustainability of U.S. economic growth since the financial crisis have been a source of anxiety and exuberance. A deep contraction in the crisis-induced recession was followed by strong — but brief — surges in the years since the crisis. In other words, after moments of respectable performance, economic growth has appeared to subside, leading policymakers and professional analysts alike to wonder whether sustainable economic growth — at longer-term, historical levels— is still possible. • The key source of U.S. economic growth is in innovation and efficiency, particularly as expressed in a constellation of technology industries and U.S. manufacturing. A trending idea in the discussion among the more apprehensive analysts of economic growth, which includes a “who’s who” of economists, senior U.S. policymakers, and market participants across the institutional spectrum, is that U.S. growth has been fundamentally impaired and that the financial crisis somehow crippled the economy for the long term. Complicating this idea of fundamental impairment are two other fears: concern over U.S. demographics and their likely impact on U.S. economic growth in the years ahead, and a belief that the United States continues to be at risk of Japan-style deflation. • Our analysis suggests that low U.S. bond yields cannot be explained by weaker long-term growth prospects. Now roughly seven years after the onset of the crisis, U.S. growth feels comfortably rapid, U.S. inflation is subdued, and Europe rather than the United States appears to be teetering on the edge of a deflation crisis. The U.S. Federal Reserve has wrapped up its quantitative easing programs, and markets are attempting to gauge the timing of the first short-term rate increase. The central uncertainty is whether U.S. economic growth will prove robust enough to mark a clean break from the era of crisis. • Aggressive monetary easing, high savings rates, and weak growth prospects outside the U.S. are depressing global interest rates; if these forces abate, we would expect bond yields to rise toward historical norms. FOR INVESTMENT PROFESSIONAL USE ONLY. NOT FOR PUBLIC DISTRIBUTION PUTNAM INVESTMENTS | putnam.com Even if the economy returns to its pre-crisis labor productivity levels, growth is still likely to be lower over the longer term. However, the yield curve could still see a significant movement upward from current levels, even with slightly lower growth weighing a bit on bond yields. In our view, there is no stagnation to fear, and consequently, the extended downward pressure on yields may soon be a thing of the past. MAY 2015 | The “old normal” returns: Forecasting long-term U.S. growth potential Growth accounting and industry-level growth analysis As part of our ongoing macroeconomic research to understand the financial building blocks related to longer-term movements in bond yields, we set out to determine our view of the long-term prospects for U.S. economic growth, and judge the merits of the case for secular stagnation. In pursuit of this inquiry, we adopted a dual methodology for developing our forecasts for growth. First, we pursued a growth accounting process that decomposes growth into its constituent sources, ranging from the growth of hours to capital investment in labor and the economy-wide impact of ongoing technological innovation. Second, we conducted an industry-level analysis of U.S. labor productivity, which we used to cross-check our growth-accounting conclusions. In contrast to concerns over fundamental stagnation, we find that a return to longer-term historical patterns is not only possible, it is likely to occur. Where does growth come from? Growth = Labor productivity + Growth of hours Growth has two major sources: labor productivity and the growth of hours worked. In this context, when we reference the growth in labor productivity or the growth of hours, we are referring to the percentage change in either of these two components. Each of these categories has a number of subcomponents, each of which possesses its own level of volatility. These sources of economic growth include demographics, improvements in the efficient use of labor and capital, the growth of capital per unit of labor, and the labor force participation rate. Growth, in other words, has a complex inner structure that responds to a variety of inputs. Some of these can be shaped and influenced by fiscal and monetary policy. Others, such as demographics, exert a kind of inexorable and immutable force. The flip side of this is easy predictability. For example, we know how many 24-year-olds there will be 24 years hence based on the current birth and mortality rates as well as emigration trends, which allows for a rare kind of precision in certain areas of economic forecasting. The horizon of the “old normal” We find the pessimism of the “secular stagnation” school to be misplaced. Projecting a long-term total real U.S. GDP growth rate of 2.2%, our analysis suggests that a return to longer-term historical growth trends is well within the realm of possibility. This result is deceptively simple — and sounds even a bit unexciting next to the quarterly GDP prints of 4.6% in Q2 2014 and 5.0% in Q3 2014. The quarterly GDP prints for the full year of 2014 illustrate our view of interim contraction and expansion of growth: Q1 -2.1%, Q2 +4.6%, Q3 +5.0%, and Q4 +2.2%. These recent GDP prints are not unlike the release of a coiled spring: When it is depressed, the spring contracts and is much smaller than when it is at rest, or in a neutral state; conversely, when the spring is released, it expands rapidly, becoming much bigger than when it is in a neutral state. But all things being equal, the spring eventually returns to its neutral state —or, in the case of GDP, it returns to its longer-run, steady state. Overall GDP growth for 2014 was 2.4%, closer to our long-run estimate, though admittedly this is merely a short-run example used to illustrate the point. Interestingly, the factors that contribute to our long-range forecast of 2.2%, particularly a factor known as “total factor productivity” (or TFP) — a measure of innovation and efficiency-enhancing elements at work in the economy — reveal an internal dynamism of U.S. growth that most observers have tended to overlook. What contributes to changes in labor productivity? %∆Capital deepening %∆Labor quality + %∆Total factor productivity = %∆ Labor productivity In our growth accounting methodology, the larger of the two factors contributing to growth is labor productivity, or the growth in output per hour. Following the economist Dale Jorgenson, a preeminent scholar of U.S. growth, we take the key components of labor productivity to be 1: •Capital deepening: Capital deepening, or the growth in capital investment per unit of labor, reflects how workers become more productive to the degree their labor is accompanied by efficiently spent capital. An example would be buying a computer for an employee whose output can benefit from such technology. In our observations, we find that capital deepening is subject to a law of diminishing returns. Investments in technology per employee, for example, tend to have natural thresholds beyond which no additional labor 1 2 Jorgenson, et al. “A Retrospective Look at the U.S. Productivity Growth Resurgence,” Journal of Economic Perspectives, Volume 22, Number 1 (Winter 2008), Pages 3–24. PUTNAM INVESTMENTS | putnam.com The volatility of labor productivity A deeper dive into labor productivity reveals diverging trends in its components over the past 35 years. While all three components of labor productivity experienced a jump in the Internet boom years, the behavior of each component on either side of that eight-year period varies considerably (Figure 2). Labor quality, for example, exhibits the characteristics of a constant, while capital deepening shows a declining trend. As we extrapolate forward in time, we expect capital deepening to converge toward lower levels, primarily because we assume the U.S. government is not likely to develop large new spending initiatives that target efficiencies in the areas of public health, education, or other social services. productivity benefit is derived from additional capital investment. If one computer increases an employee’s output, two computers may marginally improve it, too, but a third computer probably adds no value. L • abor quality: This component is defined as labor input per hour worked; it reflects changes in the composition of the workforce, as well as the difference between the growth in the compensation-weighted index of labor input and hours worked. For our study, we use labor quality statistics provided by the EU KLEMS Growth and Productivity Accounts database, a recently developed standard source of economic data containing industry-level measures of output, inputs, and productivity for 25 European countries, Japan, and the United States since 1970. In contrast to the relative consistency of labor quality and the declining trend of capital deepening, the third component of labor productivity, TFP, exhibits marked volatility. Considering the three eras of growth, TFP not only progressed from being a net detractor in the first era to being a critical positive contributor in the technological revolution beginning in 1996, but it also contracted more substantially than either of the two other components during the financial crisis. T • otal factor productivity (TFP): TFP — sometimes called multi-factor productivity or MFP — measures improvements in the efficient use of labor and capital. It includes the effects from changes in frequently unmeasured inputs, such as innovation, research and development, and other intangible investments. Replacing an employee’s computer with a more powerful machine, for example, adds to TFP. In addition, TFP quantifies the effects from organizational and institutional change, as when a firm streamlines its patterns of production and thereby realizes productivity enhancements. More than any other factor, we believe TFP has the capacity to make or break growth in the United States. But in our analysis, we see TFP heading toward being a potentially stable positive contributor to labor productivity, and therefore growth, over the long term. When we consider the components of U.S. growth since the late 1970s, three distinct eras present themselves: 1978–1995, 1996–2004, and 2005–2010. On this view, visible trends include a decline in the growth of hours and a fairly volatile contribution to growth from labor productivity (Figure 1). In addition, the 2005–2010 period starkly registers the impact of the 2008 financial crisis and the ensuing recession’s well-documented impact on employment — and, hence, the negative growth of hours. Furthermore, the immediately preceding period of 1996– 2004 just as clearly reflects the dawn of the Internet age and a massive increase in capital deepening (Figure 2). Figure 2: Volatility of TFP within labor productivity 3.0% 2.5% 2.0% 1.5% 1.0% Figure 1: Historical components of U.S. growth 0.5% 0.0% 4.0% 3.5% -0.5% 1978–1995 Labor quality 3.0% 2.5% 2.0% Source: Putnam research, 2014. 1.5% 1.0% 0.5% 0.0% -0.0% 1978–1995 Hours 1996–2004 2005–2010 Labor productivity Source: Putnam research, 2014. 3 1996–2004 Capital deepening 2005–2010 TFP MAY 2015 | The “old normal” returns: Forecasting long-term U.S. growth potential The inner workings of TFP According to our research, most of the innovation and efficiency that comprises TFP has historically come from four economic areas in the United States: retail, manufacturing, finance, and information and communication technology (ICT). Other areas of the economy, such as the government, health care, and education, have historically been negative contributors to TFP. In other words, 37% of the economy is producing most of the innovation and efficiency that can drive sustainable growth in the United States. crisis. Though it did not completely offset the declines in retail and manufacturing, ICT has proven to be, since 1996, the strongest source of innovation and efficiency in the U.S. economy. Plateauing TFP in manufacturing and retail Before we delve further into ICT to examine the key sources and trends of its contribution to growth, we think it is worth considering the fate of retail and manufacturing in the United States. Manufacturing offered a key support to TFP after 2000, but by 2004, this contribution reached a plateau. Notably, our research reveals similar trends in other developed countries. In our view, the root cause of limited growth in the United States as well as in other developedmarket manufacturing TFP can be found in globalization, which shrunk manufacturing’s share in overall output. To the degree globalization increased cross-border competition beginning in the latter half of the 1990s, it rapidly created substantial cost-cutting pressures for companies in a broad swath of industries around the world. The introduction of cost controls and labor reductions clearly took their toll on this segment of the global economy. The percentage job loss in manufacturing over the ensuing 10 years in the United States, for example, ranged from 20% to more than 40% in most states, with some of the most populous regions seeing between 30% and 40% declines. In this way, even as manufacturing continued to increasingly support TFP until 2004, manufacturing’s share of GDP declined after 2000. Within that 37%, an even smaller share of the economy is consistently delivering innovation and efficiency. When we examine industry-level data for TFP, we find that its key sources have undergone a number of important changes over the past two decades. The retail and manufacturing components, for example, almost tripled their contributions to the TFP portion of productivity from the first growth era to the Internet age, but these contributions all but disappeared in the 2005–2010 period (Figure 3). TFP in finance, moreover, appears in our analysis to be highly dependent on financial markets’ performance. When times are good, the trend is for industry players to spend more and less efficiently; when times are bad, costs are cut and labor is rationalized, giving an artificial and cyclically fragile boost to the sector’s TFP contribution. Figure 3: Total factor productivity in the U.S. is largely explained by four industries For its part, U.S. retail was a critical engine of U.S. growth at the start of 1990s, and it remained a strong contributor to TFP until approximately the year 2000. One of the drivers of this strength in the United States, we believe, was the rise of big-box retailers. Companies like Walmart were able to deploy a variety of ICT innovations, including inventory management systems and operations management software, increasing their already-powerful economies of scale. 1.5% 1.0% 0.5% 0.0% -0.5% 1978–1995 Manufacturing 1996–2004 ICT Finance R&W However, as these retail giants spread during the 1990s into densely populated areas, particularly in the northeastern United States and California, a plateau in their TFP contribution naturally took shape. The markets reached a saturation point. The top eight U.S. retailers now dominate the most densely populated areas of the country. From this point, it is increasingly difficult to raise the quotient of innovation and efficiency across their vast operation networks. 2005–2010 Others Source: Putnam research, 2014. On closer inspection, we find the true engine inside TFP to be ICT. Unsurprisingly, this area, which includes a constellation of technology industries, experienced significant growth through the first decade of the Internet age. But this area also continued to deliver strong results in the period that includes the financial 4 PUTNAM INVESTMENTS | putnam.com The keys to growth: An industry-level assessment Peeling back the layers of U.S. growth, we have found that the growth of hours worked is in decline, as the labor participation rate has yet to recover to pre-crisis levels, and that U.S. growth increasingly rests on the strength of labor productivity. Within labor productivity, we have determined that one of the largest absolute contributors, capital deepening, is in decline, and that a second factor, labor quality, has been relatively constant over the past three decades. The third element, TFP, appears to be a vital component of economic growth, but it has also proven to be the most volatile source due to large-scale macroeconomic and industry-related change, including globalization and the evolving nature of retail and wholesale activity in the United States. We have also found that the core of innovation and efficiency — and, hence, the core of TFP and the future potential of U.S. growth — appears to reside in ICT. But this, too, requires investigation. To pursue this part of our study, we conducted an industry-level analysis of labor productivity, independent of our previous growth-accounting methodology, and in this way constructed an industrial base for checking our growth forecasts. As with the prior method, labor productivity is composed of the same three components: TFP, capital deepening, and labor quality, with total labor productivity equaling the sum of the various industries we examined. In assembling this industry-by-industry view across the same three periods of growth we identified in the first part of our study, it becomes apparent that an industry’s share of GDP does not correlate with its contribution to labor productivity (Figure 4). Figure 4: The labor productivity of economic sectors does not match their share of GDP in all cases GDP share % GDP share % Social 24.4 Construction 4.3 Real estate 12.7 Transport 2.9 Retail/wholesale 11.9 Food/lodging 2.9 Professional 10.6 Electric/gas 2.1 Manufacturing 10.0 Mining 1.7 Agriculture 1.0 Finance 7.9 ICT 7.5 Labor productivity sector breakdown 12 Labor productivity (%) 10 8 1978–1995 LP 6 1996–2004 LP 4 2005–2010 LP 2 Long run LP 0 -2 -4 ICT Manufacturing Finance Retail/ wholesale Real estate Agriculture Mining Electric/ gas Source: Putnam research, 2014. 5 Construction Transport Food/ lodging Professional Social MAY 2015 | The “old normal” returns: Forecasting long-term U.S. growth potential Social spending (public administration and defense, health, education, social work, etc.), for example, has the largest GDP share — twice as much as the next-largest industry by share of GDP (real estate). And yet, although it is more than three times the size of ICT in terms of share of GDP, we calculate its long-run labor productivity to be zero percent versus ICT’s 8%. As Figure 4 shows, ICT has historically generated the largest percentage of labor productivity relative to all other industries, although volatility of this factor is high almost regardless of which industry one examines. An exception is the social spending component, which has historically remained quite flat, ranging from zero percent to 0.3% — with the positive endpoint of this range occurring during the 2005–2010 period. Interestingly, social spending in the United States is virtually always a lower contributor to labor productivity than anywhere else in the developed world (Figure 5), suggesting an area of the U.S. economy that has ample room for long-range improvement. Figure 5: U.S. social service labor productivity is lower than that of most other developed countries 3.0% 2.5% 2.0% 1.5% 1.0% 0.5% 0.0% -0.5% 1978 1981 1984 1987 1990 1993 1996 1999 2002 2005 2008 -1.0% -1.5% United States United Kingdom France Germany Japan Source: Putnam research, 2014. ICT and manufacturing: Bastions of productivity Looking across the 13 sectors in our illustration, we find the total long-range labor productivity figure to be 1.6%. Growth of hours worked, we find, will add an additional 0.6%. Thus, two thirds of our 2.2% long-range economic growth rate will be composed of labor productivity. Of that figure, we expect 0.9% — or 56% of all labor productivity — will be generated by two industries: ICT and manufacturing, which supply 0.6% and 0.3% contributions, respectively (Figure 6). Figure 6: Sector contribution to total labor productivity 0.7% 0.6% 0.5% 0.4% 0.3% 0.2% 0.1% 0.0% -0.1% ICT Manufacturing Real estate R&W Finance Transport Agriculture Source: Putnam research, 2014. 6 Professional Food Mining Electric/gas Social Construction PUTNAM INVESTMENTS | putnam.com While the four main components of ICT — electrical/optical equipment, publishing (think videos and digital music and books), telecommunications, and IT/software — have historically all played important roles, the dominating force of productivity within this area comes from electrical and optical equipment. Figure 7: Labor productivity in information and communications technology has four main components GDP share % 1978–1995 LP 1996–2004 LP 2005–2010 LP Long run LP Contribution of ICT to LP Electrical and optical equipment 1.9% 10.0% 20.5% 15.5% 18.5% 0.3% Publishing 1.5 3.5 4.7 1.4 3.4 0.1 Telecommunications 2.4 3.7 5.2 6.3 6.3 0.2 IT and others 1.7 11.0 8.9 3.8 3.0 0.1 Total ICT 7.5 6.2 9.8 7.0 8.0 0.6 Source: Putnam research, 2014. Semiconductors, computers, handheld devices, and other high-tech optics have together comprised the strongest source of labor productivity in the United States, and not just historically in the Internet’s heyday. Indeed, by our analysis, we are in the early innings of this increase in labor productivity, which we expect will converge to its 1996 trend of 18.5% — which translates into 0.3 percentage points of the total 2.2% long-term growth rate we envision for the United States. Figure 8: Labor productivity in electrical and optical equipment (moving average) 25% 20% 15% 10% 5% 0% 1978 1981 1984 1987 1990 1993 1996 1999 2002 2005 2008 Source: Putnam research, 2014. Some areas of ICT, in fact, appear to be in decline, even as optical/electrical equipment labor productivity continues to improve greatly. Labor productivity in software, for example, is feeling the drag of global outsourcing, while we expect the productivity inherent in telecommunications to converge toward a pre-financial-crisis average. What does 2.2% long-range growth mean for bond markets? Our growth accounting and industry analysis show labor productivity and total GDP growth to be progressing toward respectable levels. Even more than the 2.2% figure, we find the key drivers of growth to have displayed resiliency to 2008 and its aftermath. Although retail is slowing and finance is under regulatory pressure, productivity in ICT and manufacturing continues to drive growth on a strong and sustainable path. Most assessments of the “equilibrium” level of interest rates rely on a handful of factors: real growth, inflation expectations, and the term premium. If real growth converges more closely with our estimation of 2.2%, the Federal Reserve is able to manage inflation to meet its target of 2.0%, and there is a modestly positive term premium, then it is easy to imagine that 7 MAY 2015 | The “old normal” returns: Forecasting long-term U.S. growth potential 10-year Treasury yields would have a fair value closer to 5%. The fact that Treasury yields are so far below this level indicates there can be a large and persistent gap between market prices and “fair value.” In the case of Treasuries, this gap reflects many factors: the impact of financial regulation and incentives for financial institutions to hold the highest-quality paper; the impact of central bank behavior (recall that foreign central banks are the largest holders of Treasuries); and the fact that global interest rates are being held down by economic weakness and high rates of saving in many parts of the world. It is an important conclusion of our research that the explanation for the low level of yield is to be found in these factors; and critically, the distribution of current economic growth does not reflect a high likelihood that U.S. real growth potential has been fatally undermined by the financial crisis and its aftermath. The goal of our research was to explore the potential for growth stagnation and whether the U.S. economy could possibly return to pre-crisis levels of growth. While we do not believe that the higher levels of growth experienced pre-crisis and certainly over the past few quarters are sustainable over the long term, we do believe that respectable, longer-term levels of growth supported by “old normal” levels of labor productivity on the order of 2.2% is achievable. Putnam Retail Management | One Post Office Square | Boston, MA 02109 | putnam.com SU840 295701 7/15