The big bang: Stock market capitalisation in the long run
In recent months, stock markets and the real economy have become disconnected. While many workers and businesses are struggling with the economic effects of the COVID crisis, stock markets have rebounded close to all-time historical highs (Capelle-Blancard and Desroziers 2020). Such short-run deviations of the markets from the economy have been widespread throughout history. But over the long run, we should expect to see a strong connection between wall street and main street. Standard theory – going back to at least the famous “Kaldor facts”– tells us that stock market size should evolve in line with real activity. Furthermore, any long-run divergence between the two should be attributable to deep institutional factors that allow more firms to get listed which increase the quantity, but not the price of listed equity (Rajan and Zingales 2003). But does the theory fit the empirical facts?
In Kuvshinov and Zimmerman (2020), we introduce new annual data on stock market capitalization in 17 advanced economies between 1870 and 2015. We then use these data to trace out the long run evolution of stock market size and disentangle its drivers.
The big bang
Figure 1 shows how advanced-economy stock market capitalization has evolved over the last 145 years. During the first century of our data, market size was stable at around one-third of GDP. But in the 1980s and 1990s, we observe a sharp structural break. Stock market capitalization skyrocketed, reaching 100% of GDP and remaining at this high level thereafter.
This ‘big bang’-increase in market cap happened in every country in our sample and in all countries bar one (Belgium), it represents the single largest structural change in capitalization during this entire 145-year period. All the macro-financial shocks of the 20th century, such as the world wars, revolutions and democratic transitions, as well as the shifting monetary and financial regimes did not impact stock markets to this extent. The question, then, becomes: what drives this structural break in the data, and which forces drive the stock market more generally?
Figure 1 Stock market capitalization in advanced economies
Notes: Stock market capitalization to GDP ratio, 17 countries. The solid line and the shaded area are, respectively, the median and interquartile range of the individual country capitalization ratios in each year.
Prices or quantities?
Market capitalization is a product of quantities – the number of shares on the market – and the prices of stocks. Hence, to understanding its drivers, we decompose market cap growth into two components: net issuances (quantities) and real equity capital gains (prices). This analysis yields a stark conclusion: all of the post-1980s increase in market cap can be attributed to higher stock prices, and none to higher issuances.
The top panel of Figure 2 shows two counterfactual scenarios for the evolution of market cap after 1985. The first (red circles) shuts off the price channel, setting real equity capital gains to their pre-1985 average, while the second (blue triangles) shuts off the quantity channel in a similar manner. These can be compared to the actual evolution of market cap, the solid black line. Counterfactual 1 shows that without stock price increases, there would have been no big bang. On the contrary, if we shut down the issuance channel in counterfactual 2, stock market size today would be close to its actual levels.
The bottom panels of Figure 2 show that this dominance of price increases is also reflected in cross-country data. Before 1985, real capital gains were on average zero, and all of the long-run growth in market cap was driven by issuance. After 1985, issuances actually fell slightly while capital gains reached a level that is historically unprecedented, averaging 3.5% per year in real terms for full three decades.
Figure 2 The big bang in market cap is driven by higher stock prices, not quantities
Panel a Counterfactual evolution of market cap
Panel b Growth decomposition before 1985
Panel c Growth decomposition after 1985
Notes: Panel a: Constant capital gains counterfactual forces the real capital gains during 1985–2015 to equal the pre-1985 average. Constant net issuance counterfactual forces net issuance relative to market cap during 1985–2015 to equal the pre-1985 average. Unweighted averages of 17 countries. Panels b and c: Market cap growth is the sum of capital gains and net issuance relative to market cap. The figure displays geometric averages of market cap growth and its components before and after 1985.
What drives the stock market?
To understand this increasing disconnect between capitalization and GDP, we need to map out the deeper underlying drivers of stock valuations. We do this using a simple decomposition based on the Gordon dividend discount model, which values a stock by the present discounted value of future cash flows. In this model, higher capitalization can be driven by three factors. The first two relate to higher listed firm cash flows or profitability: either current listed-firm-profits as a share of GDP are high, or profits are expected to grow at a high rate in the future. The third factor relates to the interest rate at which expected future cash flows are discounted: the lower the discount rate, the higher the share prices.
We disentangling the drivers in two steps. First, we run predictive regressions to see if high market capitalization is correlated with either of these three factors: i.e. a high current share of listed-firm-profits to GDP, high future dividend growth, and low future returns or discount rates. We find that high capitalization correlates with high profit shares of listed firms and a low discount rate (future return), but, if anything, low future growth of dividends. This allows us to rule out an expected increase in future corporate profitability as the main source of current stock valuations. Second, we look at the time trends in the two drivers which are correlated with market cap – profit share and the discount rate. We find that both have contributed to rising stock valuations and market cap, and that the profit share channel plays the key role.
The top panel of Figure 3 shows the earnings and dividends paid by listed firms relative to GDP for our sample of 17 countries. Both earnings and dividends have increased sharply, roughly tripling since 1990 – an increase similar in magnitude to that of total market cap. The dividend data are available for the full 145-year period, and show that this recent increase is historically unprecedented.
The bottom panel of Figure 3 shows that this profit increase is essentially a distributional shift: not only have listed firms’ profits increased, but they also constitute an ever-increasing share of capital income. This echoes the findings of Eeckhout et al. (2020), who show that profitability and mark-ups of listed firms in the US have risen sharply since the 1980s, and that the distribution of profits across firms has become increasingly unequal. Another factor contributing to the increasing profit share is a redistribution of income from labour to capital (Karabounis and Nieman 2014), which is identified as a key driver of the recent market cap increase in the US by Greenwald et al. (2019).
Figure 3 Profits of listed firms have increased, while other parts of the economy have stagnated
Panel a Earnings and dividends of listed firms
Panel b Listed firm profit share in capital income
Notes: Unweighted averages of 17 countries. Profit data are aggregated up from Compustat Global and Compustat North America and cover all listed firms with non-missing values for market cap, dividends and earnings, scaled up to match our aggregate market cap data where necessary. Listed firm profits are smoothed using a 5-year moving average. Data on gross and net capital income come from Bengtsson and Waldenström (2018).
To map firm profits into stock prices, we need an estimate of the equity discount rate. This discount rate is composed of two parts: the first part is the amount investors are willing to save in general (the safe rate) and the second part is the risk they are willing to bear (the risk premium). In Kuvshinov and Zimmerman (2020b), we estimate the risky discount rate and its components again using a version of the dynamic Gordon model. This estimate of the ex-ante expected return – or discount rate – on equities is shown in Figure 4, decomposed into the safe rate and the risk premium.
The equity discount rate is currently at historically low levels. The initial fall in the discount rate in the 1970s and 1980s slightly preceded the big bang and was largely driven by a fall in the risk premium, which we link to falling macroeconomic volatility and a lower price of risk. But over the past decade and a half, risk premia have increased, and the main reason why the equity discount rate is still low is the well-documented safe rate decline.
Figure 4 Equity discount rates are at historically low levels
Panel a Equity discount rate
Panel b Ex-ante equity risk premium
Notes: The discount rate is the sum of the dividend-price ratio and expected cash flow growth. The real safe rate is the country-specific trend long-term rate estimated using the method of Del Negro et al. (2019). The risk premium is the difference between the discount rate and the safe rate. See Kuvshinov and Zimmermann (2020b) for more details on the estimation.
In summary, we trace the surge in market capitalization, and the consequent long-run disconnect between stock markets and real activity back to just two causes. First, and most importantly, even though economic growth has stagnated and future growth prospects do not look particularly vibrant, listed firms have taken on an ever-increasing share of the rewards from real economic activity, likely driven by their increasing market power vis-à-vis other firms and workers (De Loecker et al. 2020, Stansbury and Summers 2020). Second, this profit shift took place during a time of historically low discount rates, which led to a historically unprecedented increase in stock valuations. We run a counterfactual exercise which shows that at least 40% of the increase in market capitalization after 1985 is accounted for by the profit shift, and around 30% by the low discount rate, with these two factors together accounting for at least 85% of the total increase in stock market capitalization after 1985.
We also show that this long-run disconnect is fundamentally different from shorter-run deviations between markets and the economy, such as the one that we observe today. While long-run gaps are primarily driven by fundamentals such as shifting profit shares, short-run gaps are entirely attributable to changing risk appetite and discount rates. We find support for Warren Buffet’s view of stock market capitalization as “the best single measure of where valuations stand at any given moment” (Buffett and Loomis 2001). Cyclically high capitalization points to ‘hot’ and risky markets, and is typically followed by poor returns, falling stock values, and a high risk of a stock market crash. This means that even though stock valuations are close to all-time historical highs, current market risk is, in fact, anything but low.
Bengtsson, E and D Waldenström (2018), “Capital Shares and Income Inequality: Evidence from the Long Run”, Journal of Economic History 78(3): 712-743.
Buffett, W and C Loomis (2001), “Warren Buffet on the Stock Market”, Fortune Magazine, 10 December.