Xavier Gabaix

Zipf's Law for Cities: An Explanation

Zipf ’s law is a very tight constraint on the class of admissible models of local growth. It says that for most countries the size distribution of cities strikingly fits a power law: the number of cities with populations greater than S is proportional to 1/S. Suppose that, at least in the upper tail, all cities follow some proportional growth process (this appears to be verified empirically). This automatically leads their distribution to converge to Zipf ’s law.

A theory of power-law distributions in financial market fluctuations

Insights into the dynamics of a complex system are often gained by focusing on large fluctuations. For the financial system, huge databases now exist that facilitate the analysis of large fluctuations and the characterization of their statistical behaviour. Power laws appear to describe histograms of relevant financial fluctuations, such as fluctuations in stock price, trading volume and the number of trades. Surprisingly, the exponents that characterize these power laws are similar for different types and sizes of markets, for different market trends and even for different countries—suggesting that a generic theoretical basis may underlie these phenomena. Here we propose a model, based on a plausible set of assumptions, which provides an explanation for these empirical power laws. Our model is based on the hypothesis that large movements in stock market activity arise from the trades of large participants. Starting from an empirical characterization of the size distribution of those large market participants (mutual funds), we show that the power laws observed in financial data arise when the trading behaviour is performed in an optimal way. Our model additionally explains certain striking empirical regularities that describe the relationship between large fluctuations in prices, trading volume and the number of trades.

Power Laws in Economics and Finance

A power law is the form taken by a large number of surprising empirical regularities in economics and finance. This article surveys well-documented empirical power laws concerning income and wealth, the size of cities and firms, stock market returns, trading volume, international trade, and executive pay. It reviews detail-independent theoretical motivations that make sharp predictions concerning the existence and coefficients of power laws, without requiring delicate tuning of model parameters. These theoretical mechanisms include random growth, optimization, and the economics of superstars coupled with extreme value theory. Some of the empirical regularities currently lack an appropriate explanation. This article highlights these open areas for future research.

The evolution of city size distributions

We review the accumulated knowledge on city size distributions and determinants of urban growth. This topic is of interest because of a number of key stylized facts, including notably Zipf’s law for cities (which states that the number of cities of size greater than S is proportional to 1/S) and the importance of urban primacy. We first review the empirical evidence on the upper tail of city size distribution. We offer a novel discussion of the important econometric issues in the characterization of the distribution. We then discuss the theories that have been advanced to explain the approximate constancy of the distribution across very different economic and social systems, emphasizing both bare-bone statistical theories and more developed economic theories. We discuss the more recent work on the determinants of urban growth and, in particular, growth regressions, economic explanations of city size distributions other than Gibrat’s law, consequences of major shocks (quasi natural experiments), and the dynamics of U.S. urban evolution.

Rank-1/2: A Simple Way to Improve the Ols Estimation of Tail Exponents

Despite the availability of more sophisticated methods, a popular way to estimate a Pareto exponent is still to run an OLS regression: log(Rank) = a − b log(Size), and take b as an estimate of the Pareto exponent. The reason for this popularity is arguably the simplicity and robustness of this method. Unfortunately, this procedure is strongly biased in small samples. We provide a simple practical remedy for this bias, and propose that, if one wants to use an OLS regression, one should use the Rank −1 / 2, and run log(Rank − 1 / 2) = a − b log(Size). The shift of 1 / 2 is optimal, and reduces the bias to a leading order. The standard error on the Pareto exponent ζ is not the OLS standard error, but is asymptotically (2 / n)1 / 2ζ. Numerical results demonstrate the advantage of the proposed approach over the standard OLS estimation procedures and indicate that it performs well under dependent heavy-tailed processes exhibiting deviations from power laws. The estimation procedures considered are illustrated using an empirical application to Zipf’s law for the United States city size distribution.

The 6D Bias and the Equity-Premium Puzzle

If decision costs lead agents to update consumption every D periods, then econometricians will find an anomalously low correlation between equity returns and consumption growth (Lynch 1996). We analytically characterize the dynamic properties of an economy composed of consumers who have such delayed updating. In our setting, an econometrician using an Euler equation procedure would infer a coefficient of relative risk aversion biased up by a factor of 6D. Hence with quarterly data, if agents adjust their consumption every D =4 quarters, the imputed coefficient of relative risk aversion will be 24 times greater than the true value. High levels of risk aversion implied by the equity premium and violations of the Hansen-Jagannathan bounds cease to be puzzles. The neoclassical model with delayed adjustment explains the consumption behavior of shareholders. Once limited participation is taken into account, the model matches most properties of aggregate consumption and equity returns, including new evidence that the covariance between ln(Ct+h/Ct) and Rt+1 slowly rises with h.

Is CEO Pay Really Inefficient? A Survey of New Optimal Contracting Theories

Bebchuk and Fried (2004) argue that executive compensation is set by CEOs themselves rather than boards on behalf of shareholders, since many features of observed pay packages may appear inconsistent with standard optimal contracting theories. However, it may be that simple models do not capture several complexities of real-life settings. This article surveys recent theories that extend traditional frameworks to incorporate these dimensions, and show that the above features can be fully consistent with efficiency. For example, optimal contracting theories can explain the recent rapid increase in pay, the low level of incentives and their negative scaling with firm size, pay-for-luck, the widespread use of options (as opposed to stock), severance pay and debt compensation, and the insensitivity of incentives to risk.

Economic fluctuations and anomalous diffusion

We quantify the relation between trading activity - measured by the number of transactions N(Deltat)-and the price change G(Deltat) for a given stock, over a time interval [t, t+Deltat]. To this end, we analyze a database documenting every transaction for 1000 U.S. stocks for the two-year period 1994-1995. We find that price movements are equivalent to a complex variant of classic diffusion, where the diffusion constant fluctuates drastically in time. We relate the analog for stock price fluctuations of the diffusion constant-known in economics as the volatility-to two microscopic quantities: (i) the number of transactions N(Deltat) in Deltat, which is the analog of the number of collisions and (ii) the variance W(2)(Deltat) of the price changes for all transactions in Deltat, which is the analog of the local mean square displacement between collisions. Our results are consistent with the interpretation that the power-law tails of P(G(Deltat)) are due to P(W(Deltat)), and the long-range correlations in |G(Deltat)| are due to N(Deltat).

Scaling and correlation in financial time series

We discuss the results of three recent phenomenological studies focussed on understanding the distinctive statistical properties of financial time series – (i) The probability distribution of stock price fluctuations: Stock price fluctuations occur in all magnitudes, in analogy to earthquakes – from tiny fluctuations to very drastic events, such as the crash of 19 October 1987, sometimes referred to as “Black Monday”. The distribution of price fluctuations decays with a power-law tail well outside the Levy stable regime and describes fluctuations that differ by as much as 8 orders of magnitude. In addition, this distribution preserves its functional form for fluctuations on time scales that differ by 3 orders of magnitude, from 1 min up to approximately 10 days. (ii) Correlations in financial time series: While price fluctuations themselves have rapidly decaying correlations, the magnitude of fluctuations measured by either the absolute value or the square of the price fluctuations has correlations that decay as a power-law, persisting for several months. (iii) Volatility and trading activity: We quantify the relation between trading activity – measured by the number of transactions NΔt – and the price change GΔt for a given stock, over a time interval [t,t+Δt]. We find that NΔt displays long-range power-law correlations in time, which leads to the interpretation that the long-range correlations previously found for |GΔt| are connected to those of NΔt.

Learning in the Credit Card Market

Agents with more experience make better choices. We measure learning dynamics using a panel with four million monthly credit card statements. We study add-on fees, specifically cash advance, late payment, and overlimit fees. New credit card accounts generate fee payments of