

H index
Stands for HerfindahlHirschman index, which is a way of measuring the concentration of market share held by particular suppliers in a market. It is the sum of squares of the percentages of the market shares held by the firms in a market. If there is a monopoly  one firm with all sales, the H index is 10000. If there is perfect competition, with an infinite number of firms with nearzero market share each, the H index is approximately zero. Other industry structures will have H indices between zero and 10000. Tirole's version is bounded between zero and one because each of the market shares is between zero and one.

Habakkuk thesis
That high wages and labor scarcity stimulated technological progress in the U.S. in the 1800s, and in particular brought about the American system of manufacturing based on interchangeable parts. (This description from Mokyr, 1990; idea from Habakkuk, 1962).

Habit
Generally, habits are conceptualized as the learning of sequences of acts that have become
automatic responses to specific situations, which may be functional in order to achieve a
given result, or to obtain certain goals or end states (e.g.
James, 1890,
Triandis, 1977,
Watson, 1914).
Habits thus comprise a goal directed type of automaticity; they are instigated by a
specific goal directed state of mind on the presence of triggering stimulus cues, for instance
taking the car to travel to the supermarket. Once evoked, the behavior will run to completion
without the need for attentional control of the process. Habit strenght is proposed to increase
as a result of repetitions of positive reinforcements.

Hahn problem
Hahn (1965) question: when does there exist an equilibrium in a model in which money has positive value?

Hansen's J test
See J statistic

Harrodneutral
A synonym for laboraugmenting, in practice.

Hausman test
Given a model and data in which fixed effects estimation would be appropriate, a Hausman test tests whether random effects estimation would be almost as good. In a fixedeffects kind of case, the Hausman test is a test of H_{0}: that random effects would be consistent and efficient, versus H_{1}: that random effects would be inconsistent. (Note that fixed effects would certainly be consistent.) The result of the test is a vector of dimension k (dim(b)) which will be distributed chisquare(k). So if the Hausman test statistic is large, one must use FE. If the statistic is small, one may get away with RE.

hazard rate
escape rate; rate of transition out of current state

Heaviside function
Is a mapping from the real line to {0, 1}, denoted (at least sometimes) hv(x). hv(x) is zero for x<0, and is one for x>=0.

Heckit
An occasional name for generalized Tobit. This approach allows a different set of explanatory variables to predict the binary choice from those which predict the continuous choice. (The data environment is one in which the continuous choice is measured only when the binary choice is nonzero  e.g., if we have data on people, whether they bought a car, and how expensive it was, we can estimate a statistical model of how expensive a car other people would buy, but only on the basis of the ones who did buy a car in the data sample.) A regular, nongeneralized Tobit constrains the two sets of variables to be the same, and the signs of their effects to be the same in the two estimated equations. 'Heck' is for James Heckman.
 Christopher Baum, Boston College economics department, 20 May 2000, in a broadcast to the statalist, the email list of people interested in the software Stata.

Heckman twostep estimation
A way of estimating treatment effects when the treated sample is selfselected and so the effects of the treatment are confounded with the population that chose it because they expected it would help  the classic example is that college educations may be selected by those most likely to benefit.
Taking that example, we wish to advance past the following regression: w_{i} = a + bX_{i} + dC_{i} + e_{i} where i indexes people, w_{i} is the wage (or other outcome variable) for agent i, X_{i} are variables predicting i's wage, and C_{i} is 1 if i went to college and 0 if not. e_{i} is the remaining error after least squares estimation of a, b, and d.

HeckscherOhlin model
A model of the effects of international trade. "The HeckscherOhlin framework typically is presented as a twocountry, twogood, twofactor model. The two countries are assumed to share identical, homothetic tastes for the two substitutable goods and identical, constantreturnstoscale technologies with some factor substitutability. Perfect competition prevails in each market with zero transport costs and no artificial barriers to international trade in goods, although factors are internationally immobile. In this framework, each country will (incompletely) specialize in production and export the good using intensively in production the factor that the country has in relative abundance." That effect is called factorprice equalization across countries, and is used sometimes to explain how rising international trade would lead to greater income inequality in the most developed countries. (from Bergstrand, Cosimano, Houck, and Sheehan, 1994, p 3) The reference in the name is to "Scandinavian economists Eli Heckscher and Bertil Ohlin early in [the twentieth century]" in work that is rarely cited directly. (from Bluestone, 1994, p 336).

Hedge strategies
A hedge strategy is the intentional reduction of the loss risk (downside risk) of an
underlying asset to the debit of the gain chance. Usually hedge strategies are done with
derivative securities, e.g. options.

hedonic
of or relating to utility. (Literally, pleasurerelated.) A hedonic econometric model is one where the independent variables are related to quality; e.g. the quality of a product that one might buy or the quality of a job one might take.
A hedonic model of wages might correspond to the idea that there are compensating differentials  that workers would get higher wages for jobs that were more unpleasant.
"A product that meets several needs, or has a variety of features ... generates a number of hedonic services. Each one of these services can be thought of as generating its own demand, along with a resulting hedonic price. Although each separate component is not observable, the aggregation of all the components results in the observed product demand and equilibrium price.... [Q]uality improvements will appear to an observer as an outward shift of the product demand curve, as consumers are willing to purchase more at the prevailing price."  William J. White, "A Hedonic Index of Farm Tractor Prices: 19101955", Ohio State University working paper, October 1998, pp. 34.

help
A list of fields contained here is below. There is some other advice at this help page: http://econterms.com/help.html
Most terms are in one of these categories. You can click on one to see a list of terms relevant to it. fields

HerfindahlHirschman index
See 'H index'.

Hermite polynomials
The Hermite polynomials are a series of polynomials defined for each natural number r, used for statistical approximations I believe. Click here for the equation and graphs of the first several.

Hessian
The matrix of second derivatives of a multivariate function. That is, the gradient of the gradient of a function. Properties of the Hessian matrix at an optimum of differentiable function are relevant in many places in economics: 1) In maximum likelihood estimation, the information matrix is (1) times the Hessian.

heterogeneous process
A stochastic process is heterogeneous if it is not identically distributed every period.

heteroscedastic
An alternate spelling of heteroskedastic. McCulloch (1985) argues that the spelling with the k is preferred, on the basis of the pronunciation and etymology (Greek not French derivation) of the term.

heteroskedastic
An adjective describing a data sample or datagenerating process in which the errors are drawn from different distributions for different values of the independent variables. Most commonly this takes the form of changes in variance with the magnitude of X. That is, in y = Xb + e that the e's vary in magnitude with the X's. (An example is that variance of income across individuals is systematically higher for higher income individuals.) If the errors are drawn from different distributions, or if higher moments of the error distributions vary systematically, these are also forms of heteroskedasticity.

Heuristic
A heuristic is a strategy that can be applied to a variety of problems and that usually ?
but not always ? yields a correct solution. People often use heuristics (or shortcuts) that
reduce complex problem solving to more simple judgmental operations. Three of the most popular
heuristics are discussed by Tversky and Kahnemann (1974):

HicksKaldor criterion
For whether a costbenefit analysis supports a public project. The criterion is that the gainers from the project could in principle compensate the losers. That is, that total gains from the project exceed the losses. The criterion does not go so far as the Pareto criterion, according to which the gainers would in fact have to compensate the losers.

Hicksneutral
An attribute of an effectiveness variable in a production function. The attribute is that it does not affect labor differently from the way it affects capital.
The canonical example is the Solow model production function Y=AF(K,L). There Y is output, L labor, K capital, F a production variable, and A represents some kinds of effectiveness variable. In Y=F(AK,L) the effectiveness variable affects capital but not labor. In Y=F(K,AL) it affects labor but not capital. These two cases can be described as Hicksbiased. In Y=AF(K,L) it is Hicksneutral.

Hicksneutral technical change
Given a production function AF(K,L) changes in A are Hicksneutral, meaning that they do not affect the optimal choice of K or L. The subject comes up in practice only for aggregate production functions.
Uzawa, H. 'Neutral Inventions and the Stability of Growth Equilibrium,' The Review of Economic Studies 28:2 (Feb., 1961), 117124 contains the first known published use of the adjective 'Harrod neutral' According to it, the criterion of Harrodneutrality comes from
Harrod, Roy F., 'Review of Joan Robinson's Essays in the Theory of Employment,' Economic Journal, vol. 47 (1937), 326330.
Uzawa also proves that AF(K,L) and F(K,AL) are the right functional forms to meet Hicks and Harrodneutrality, and that only the CobbDouglas form accomplishes both.

Hicksian demand function
h(p,u)  the amount of a good that demanded by a consumer given that it costs p per unit and that the consumer will have utility u from all goods. h(p,u) is the costminimizing amount.

High School and Beyond
A panel data set on U.S. high school students.

highpowered money
reserves plus currency

Hilbert space
A complete normed metric space with an inner product. So the Hilbert spaces are also Banach spaces. L^{2} is an example of a Hilbert space. Any R^{n} with n finite is another.

Hindsight bias
It is a common observation that events in the past appear simple, comprehensible, and
predictable in comparison to events in the future. Everyone has had the experience of
believing that they knew all along the outcome of a football game, a political
election or a business investment. The hindsight bias is the tendency for people with
outcome knowledge to believe falsely that they would have predicted the reported outcome
of an event. After learning of the occurrence of an event, people tend to exaggerate the
extent to which they had foreseen the likelihood of its occurrence.

Hindsight bias biased reconstruction
Contemporary models suggest that the hindsight bias appears due to reconstructive processes while people try to generate their original estimates.
One is loosely based on the
response bias hypothesis,
originally developed in eyewitness testimony research: People are assumed either to have remembered or to have forgotten their original judgements. Those that do remember their original estimates are likely to reproduce them. Those who have forgotten them are forced to guess and, in the presence of outcome information, are likely to utilize this information as an anchor assuming that their estimates must have been somewhere in the proximity of the true outcome. But since people are generally optimistic about their capacities, they will locate their presumed prior estimates closer to the real outcome than it had actually been, resulting in the hindsight bias.
Other authors concentrate their assumptions on three different stages in the reconstructive process: Selective retrieval, prejudiced interpretation and weighing of different cues.
Finally, the hindsight distortion may be triggered by the heuristic of
anchoring and adjustment.

Hindsight bias implications for further research
Hindsight Bias is a strong phenomenon which has been observed in many circumstances,
and appears to influence everyday decisionmaking. Researchers are trying to develop
hindsight bias models to provide a better
theoretical explanation.
Another aspect is
practical relevance,
whereby reallife cases need to be examined for evidence on distortion influence.

Hindsight bias memory impairment
Fischhoff´s original explanation of the hindsight bias, Immediate Assimilation Hypothesis,
states that memory for original predictions is altered by subsequent outcome
(Fischhoff, 1975). When learning
about the actual or alleged outcome, the person reinterprets the original evidence in the
light of the outcome. They are therefore inadvertently modifying what had been previously
stored in memory. Subsequent outcome knowledge is integrated immediately into the existing
knowledge structure. This results in a permanent modification of the person´s prior
representation of the event. Other variations of the memory impairment hypothesis
suggest that the origins of hindsight biases lay in the retrieval stage. The Selective
Retrieval Hypothesis maintains that known outcome serves as a retrieval cue for relevant
case material. Once an outcome has been learned, information congruent with this outcome
will become highly accessible. Incongruent information cannot be retrieved with the same
ease. The authors of the Dual Memory Traces Model
(Hell, Gigerenzer, Gauggel, Mall & Müller, 1988)
suggested an extension of Fischhoff´s model. They assume two separate memory traces for own
judgements and subsequent outcome information. The strength of hindsight biases is determined
by the relative strength of the memory traces.

Hindsight bias motivational explanations
Motivational explanations suggest that judgement and decision processes are not only
affected by rational cognitions. They are also influenced by actual needs and motives,
including need for control, need for cognition, selfrelevance and,
most importantly selfpresentation concerns.
In the latter approach, people are motivated to make others
believe that their predictions were close to the actual outcome, in an attempt to
maintain a high level of public selfesteem. Contrary to the memory impairment
hypothesis, this explanation interprets hindsight distortions as adjustments during
the response generation state. Several authors showed, that empirical evidence for
motivational underpinnings of the knewitallalong effect is rather weak.

Hindsight bias practical relevance
Everyday life provides numerous examples on the practical relevance of hindsight bias.
This was shown in a medical context. A GP´s second opinion does not differ completely from
another GP´s opinion, if he/she is aware of the first opinion. This seems to be of serious
consequence if one considers that a second opinion is only required when serious illnesses
have been diagnosed.
In a legal context, hindsight bias was found to occur when a jury makes a final decision
in court. In the course of a trial, the judge is empowered to order the jury to ignore
certain testimonials, by disallowing them. It has been proven impossible to ignore such
information.
Hindsight bias also plays a role in an economic context. An economic expert may, for
example, analyze certain share activity, as if he/she had always known what would happen.
This results in them forming a higher opinion of their own judgement. This in turn can have
a longterm, restrictive effect on their individual learning capability as well as future
decisionmaking. In addition to this intrapersonal effect, it is possible for interpersonal
effects to arise. Example: A supervisor may no longer be able to make an undistorted judgement
on his employees´ decisionmaking if he/she got information about some results of their
performance. This is a special problem in the case of poor outcome because it could happen,
despite them having acted correctly on the information they were given at the time.

Hindsight bias response bias hyothesis
This model was originally conceived during eyewitness testimony research. This was to account for the fact that witnesses receiving misleading information about a previously observed event, show a poorer memory for it. Similar to the hindsight bias, the misleading information effect was originally attributed to memory impairment.
Subsequent interpretations offered the following explanation:
Rather than altering existing memory traces, the new information may be used as a reference point by those who cannot remember the original information and therefore need to guess. Misleading information does not alter the original representation, but simply serves as an anchor to perceivers who are unable to retrieve it.
Parallels between the hindsight bias and the misleading information paradigms are obvious. Both lines of research query whether information stored in the memory might be less accessible after being confronted with inconsistent new information. The experimental designs of both research traditions show strong similarities.
Important differences: In the misleading information paradigm, the original information is presented by the experimenter. In hindsight bias studies however, the original estimates are generated by the subjects. Misleading information is presented unobtrusively without the subjects being aware of its misleading nature. Outcome information given in hindsight studies is explicitly labeled as the correct information.

Hindsight bias theoretical and empirical work
More recent studies have come up with models that enable precise forecasting of the strength and direction of the hindsight distortion, using a quantitative basis. Researchers are also attempting to determine how far hindsight decisions affect a person´s trust in their own judgements. They are also trying to determine how far such decisions affect a person´s impression of their competency in future decisionmaking on the same subject. Another interesting point to be examined is the relationship between hindsight bias and the attribution of responsibility.

Hindsight bias theoretical explanations
Although there is a wealth of literature on hindsight distortions, the underlying
mechanisms are not yet fully understood. Attempts of explanation were given in three major
theoretical areas: Models of
memory impairment
state that outcome knowledge affects memory for previous judgements by either altering
or erasing existing memory traces, or by rendering them less accessible. Other attempts
to explain the hindsight bias are based on assumptions that these distortions are driven by
motivations.
An alternative explanation is that people use distorting heuristics while
reconstructing
their original judgements.

history
The subject of economic history is anything in history that is subject to economic explanations. Application of formal theory or statistical analysis of data may be relevant, although it is possible to make a contribution without either, e.g. with a case study or a contextual reinterpretation. Historians tend to be focused on what happened, how, and why, not on the question of whether a model fits the evidence. history

HLM
Statistical software for Hierarchical Linear Modeling, from Scientific Software International.

HodrickPrescott filter
Algorithm for choosing smoothed values for a time series. The HP filter chooses smooth values {s_{t}} for the series {x_{t}} of T elements (t=1 to T) that solve the following minimization problem: min { {(x_{t}s_{t})^{2} ... etc. } Parameter l>0 is the penalty on variation, where variation is measured by the average squared second difference. A larger value of l makes the resulting {s_{t}} series smoother; less highfrequency noise. The commonly applied value of l is 1600. For the study of business cycles one uses not the smoothed series, but the jagged series of residuals from it. See Cooley, 1995, p 2729. That HP filtered data shows less fluctuation than firstdifferenced data, since the HP filter pays less attention to high frequency movements. HP filtered data also shows more serial correlation than firstdifferenced data. For l=1600: "if the series were stationary, then [this choice] would eliminate fluctuations at frequencies lower than about thirtytwo quarters, or eight years."

holdup problem
One of a certain class of contracting problems.
Imagine a situation where there is profit to be made if agents A and B work together, so they consider an agreement to do so after A buys the necessary equipment. The holdup problem (in this context) is A might not be willing to take that agreement, even though the outcome would be Pareto efficient, because after A has made that investment, B would have the power might decide to demand a larger share of the profits than before, since A is now deeply invested in the project but B is not, so B has some bargaining power that wasn't there before the investment. B could demand all of the profits, in fact, since A's alternative is to lose the investment entirely.
Other holdup problems are analogous to this one.

Holder continuous
An attribute of a function g:R^{d}>R. g can be said to be Holder continuous if there exist constants C and 0<=E<=1 such that for all u and v in R^{d}: g(u)g(v) <= Cuv^{E}
So if g is Holder continuous for C=1 then it is Lipschitz continuous? And if g is Holder continuous then it is continuous.

homoscedastic
An alternate spelling of homoskedastic. McCulloch (1985) argues that the spelling with the k is preferred, on the basis of the pronunciation and etymology (Greek not French derivation) of the term.

homoskedastic
An adjective describing a statistical model in which the errors are drawn from the same distribution for all values of the independent variables. Contrast heteroskedastic. This is a strong assumption, and includes in particular the assumption in a linear regression, for example, y = Xb + e that the variance of the e's is the same for all X's.
(The observed variance will differ in almost any sample. But if one believes that the datagenerating process does not in principle have greater variances for different values of the independent variable, one would describe the sample as homoskedastic anyway.)

homothetic
Let u(x) be a function homogeneous of degree one in x. Let g(y) be a function of one argument that is monotonically increasing in y. Then u(g()) is a homothetic function of y.
So a function is homothetic in y if it can be decomposed into an inner function that is monotonically increasing in y and an outer function that is homogeneous of degree one in its argument.
In consumer theory there are some useful analytic results that can come from studing homothetic utility functions of consumption.

Household behavior
In traditional microeconomics, household behavior is understood
narrowly as the theory of consumer demand for commodities, i.e., household
consumption. There are, however, other aspects of household
behavior that have also been investigated in the microeconomics literature, such as
the household´s supply of labor, the production of commodities (mainly, services) within
the household (household production), saving decisions,
retirement decisions, and many more.

HRS
Health and Retirement Study, a longitudinal panel of older Americans studied by the Survey Research Center at the University of Michigan. Their Web site is at http://www.umich.edu/~hrswww.

HSB
High School and Beyond, a panel data set on U.S. high school students.

Huber standard errors
Same as HuberWhite standard errors.

HuberWhite standard errors
Standard errors which have been adjusted for specified assumedandestimated correlations of error terms across observations.
The implicit citations are to Huber, 1967, White, 1980, and White, 1982.

human capital
The attributes of a person that are productive in some economic context. Often refers to formal educational attainment, with the implication that education is investment whose returns are in the form of wage, salary, or other compensation. These are normally measured and conceived of as private returns to the individual but can also be social returns.
''Human capital' was invented by the economist Theodore Schultz in 1960 to refer to all those human capacities, developed by education, that can be used productively  the capacity to deal in abstractions, to recognize and adhere to rules, to use language at a high level. Human capital, like other forms of capital, accumulates over generations; it is a thing that parents 'give' to their children through their upbringing, and that children then successfully deploy in school, allowing them to bequeath more human capital to their own children.'  Traub (2000)

hyperbolic discounting
A way of accounting in a model for the difference in the preferences an agent has over consumption now versus consumption in the future.
For a and g scalar real parameters greater than zero, under hyperbolic discounting events t periods in the future are discounted by the factor (1+at)^{(g/a)}.
That expression describes the "class of generalized hyperbolas". This formulation comes from a 1999 working paper of C. Harris and D. Laibson, which cites Ainslie (1992) and Loewenstein and Prelec (1992).
In dynamic models it is common to use the more convenient assumption that agents have a common discount rate applying for any tperiod forecast, starting now or starting in the future. Hyperbolic discounting is less convenient but fits the psychological evidence better, and when contrasted to the constantdiscountrate assumption can get models to fit the noticeable fall in consumption that U.S. workers are observed to experience when they retire. In a constantdiscountrate model the worker would usually have forecast the fall in income and their consumption expenses would be smooth.
One reason hyperbolic preferences are less convenient in a model is not only that there are more parameters but that the agent's decisions are not timeconsistent as they are with a constant discount rate. That is, when planning for time two (two periods ahead) the agent might prepare for what looks like the optimal consumption path as seen from time zero; but at time two his preferences would be different.
Contrast quasihyperbolic discounting.

hysteresis
a hypothesized property of unemployment rates  that there is a ratcheting effect, so a shortterm rise in unemployment rates tends to persist. Theories that would lead to hysteresis:  an insider/outsider model of decisionmaking about employment; insiders such as the unionized workers ratchet up wage rates beyond where it is profitable to hire the unemployed; outsiders who are unemployed don't get to be part of the negotiation process.  behavioral and human capital changes among the unemployed, such as forgetting the details of work or work behavior, or losing interest or skill in getting new jobs, could lead to declining chances of becoming employed.
