Initially, the VaR has been anticipating to quantify the available risks in derivatives markets, but it has grown widely and it has now been applied in measuring all kinds of risks, primarily credit and market risks. It also developed from a tool that quantifies risk to a tool that is applied in active risk management. Today VaR has shifted beyond application in financial institutions. In the beginning, companies with largely exposed to financial markets used other kinds of activities before spreading to other businesses. Today, an ever-growing numbers of individual businesses apply and appreciate VaR as an effective tool for quantifying financial risksKrause (2003). This trend is evidently aided by the fact that non-specialists easily understand VaR. The risks of the prevalent use of VaR are an overdependence on the results it gives, misconception, and even abuse. It is as a result that, essential individuals using VaR understand its problems and limitations. In this paper, I will explore in depth these constraints, which unluckily do not mark prominently.

To begin with, the VaR estimate is founded on precedent data, that is, it uses past distribution of effects of the investment. However, to calculate the peril of an investment, it is of no concern how big this risk has been in the earlier period, but fairly on how much exposure there is within the existing period; therefore, the future distribution of outputs would be the appropriate to consider. As long as the division of outcomes stays steady, the VaR can simply be removed from the past distribution. In reality, the distributions are not steady over time; most remarkably, the inconsistency of outcomes and the correlations vary. Relying exclusively on past data can as a result give a poor risk measure Oldfield, et al (2000).

Unluckily, even with this problem resolved, there lingers an issue with the evaluation of the VaR itself. Since the true probability circulation is not well known in general, it has to be evaluated from the data. A good evaluation of the minor tail of the distribution is of crucial importance for the VaR evaluation. The obstacle comes from the fact that from the definition, only few remarks are made at the tails. This enlarges the estimation errorsKrause (2003).

On the other hand, VaR applies to several financial instruments and it is stated in similar unit of quantifying, that is, in “lost money”. On the converse, Greeks are measures created ad hoc for precise instruments or risk variables and are stated in deterrent units. The contrast of comparative riskiness between, say, a forex portfolio, and an equity portfolio is not simple with Greeks, whilst it is a straight contrast knowing their VaR’s. Secondly, VaR includes an educated guess of potential events and allows one to change in a solitary number the risk of a collection. On the contrary, Greeks essentially add up to “what if” variables Acerbi, et al (2008).

In real scenarios, due to the intricacy of computational features and to the fragility of the approximation of market possibilities, in order to calculate VaR one has to choose (sometimes strong) hypothesis both on the practical reliance of financial instruments from peril drivers (first and second order estimates . . .) and on the distributions (past VaR, parametric VaR . . .). Sometimes it is practical to decouple the risks related to diverse risk drivers. VaR can be then calculated “switching on” some category of risk drivers, keeping all the remaining constant. One then talks of partial VaR’s like “Forex VaR” (FXVaR), “Interest Rate VaR” (IRVaR), and “Credit VaR” (CVaR)

(EQVaR), “Equity VaR”, and so onAcerbi, et al (2008).

In the case of composite portfolios uncovered to many risk erratic like in financial institutions, the calculation of VaR can frequently be an alarming task. A demanding feature is due to a fact that the calculation cannot be divided into different sub–calculations because the two-fold non-additivity property of VaR. In certain cases of normal distribution returns of a portfolio, an individual can show the “non–additivity” is a “sub–additivity”. Total VaR is constantly equal or less than the sum of partial VaR’s Oldfield, et al (2000).

VaR is in fact receptive to the hedging impact of deterrent positions and the mutual correlation impact of risk drivers. The sub–additivity in VaR of a Gaussian environment embodies the familiar belief that varying lower risks. We are now in a point to appreciate the most recurrent criticisms used against VaR, which in the earlier period gave rise to cruel battles linking opposite factionsAcerbi, et al (2008) .“ VaR constantly come late when the harm is already done”: this adage comes from the fact that for one to approximate market probabilities it is a universal habit to attune potential scenarios on historical market data. For instance, it is apparent that the day prior to some market unrest, the parameter guesses will not be able to project the sudden hops of volatilities hence VaR will unavoidably miscalculate the risks, identifying the rise in risks some days later.

However, this objection should not be tackled to VaR itself, but rather to a sure (and widespread) way of calculated VaR by approximating potential probabilities based on historical data. For example, in a trading desk this could be shun by requiring correct volatility contributions from the traders who can timely forecast future smacks of volatility from market fundamentals or from the news. “VaR does not sense”: it may appear pessimism, but VaR approximates, particularly in the case of composite portfolios may be a tricky computational duty that the last figure may lack statistical value. It is not impossible in these cases to see illogical VaR numbers that do not relate to the risks in the portfolio.

The rest of assumptions, statistical errors, and estimates introduced in the calculation can be so huge that the result is nonsense because of its widespread confidence interval. These inspections can seem pessimistic. However, for any other statistical guess the right thing to do is: one should not belief in it except if all the assumptions and estimates made for the approximation are under scrutiny and except if the user of the outcome is aware of the guesses that have been made.

VaR is not easy to use, it fails to be a coherent risk measure, and its estimation is subject to large errors. Individuals within the company as well as the company as a whole can utilize these limitations. However, these limitations do not mean that VaR is not a valuable tool in risk management. The apparent advantage of VaR is that non-specialists simply and instinctively understand it. Therefore, it can be well conversed within a company as well as, stakeholders, or other investors Krause (2003). Furthermore, it can address a multiple types of risks in a solitary framework, which permits the aggregation of risks in addition to facilitating communication. With the aid of VaR, we can address most issues involving risks. It can be applied in setting risk bounds for the entire company, divisions, and traders. Next, it assists routine measurement and may be used for routine-related pay. Thirdly, it facilitates conclusions on the distribution of capital and gives a sign of the total capital prerequisite. Finally, it helps decide which risks to lessen, if necessary.

No other risk management tool developed thus far tackles all these features in a single framework, while it is still available to directors and a wide range of staff. Therefore, VaR has established to be a useful tool that has eagerly accepted, despite its limitations.

The VaR approximate must not be considered as an exact number, but it offers a direction as to how the extent of risk that has been engaged. It also aids in noticing any movements in the mannerisms of individuals, or t companies as a wholeKrause (2003). When properly used, VaR is a dominant and simple tool when managing risks. On the contrary, overdependence on its outcomes and mitigating important choices solely on its basis is probably counterproductive. No risk-managing tool can substitute the sound ruling of managers, those applying it should be conscious of its limits.