You are on page 1of 13

ASSET PRICING, INTEREST RATES, EXPECTED INFLATION, AND MONETARY POLICY

Asset Pricing How is the price of an asset (i.e. a house, a companys stock, etc.) determined? To answer this question we will study an asset pricing model due to Lucas [1978]. The model is a minor modication to the standard model we have been using, augmented only in that we will allow for uncertainty. As such, the model is extremely simple, yet key in understanding the forces behind the determination of asset prices. The model is a simple general equilibrium model, in which to avoid unnecessary complications we will abstract from production (i.e. we will work in an endowment economy). The economy consists of a great deal of individuals who interact with each other through nancial markets. In the economy, the asset will be a tree which each period confers to its owner a random amount of fruit dt (for example, dt can be taken as dividends if the asset is a companys stock).1 At each point in time agents can buy or sell trees, but we assume that the total number of available trees is xed and is equal to the number of agents in the economy. We will assume that the fruit dt cannot be stored, that is, it is either consumed at t or it is lost forever. Let at denote the number of trees possessed by an individual at time t, and pt the price of trees at date t. In addition to the purchase of trees, consumers devote part of their resources to purchases of consumption goods, therefore, an individuals budget constraint is given by ct + pt at+1 (pt + dt ) at . The right-hand side of this budget constraint shows that the agent receives income from two sources: 1. from the dividends paid by the trees he owns, and 2. by the sale of the trees he owns. The agents utility maximization problem is given by t max E0 u (ct )
{ct ,at+1 }t=0 t=0

s.t. ct + pt at+1 (pt + dt ) at

Here, E [] denotes mathematical expectation, and it makes explicit that future dividends are uncertain and the agent must take this into account when formulating his
1This model is sometimes referred to as the Lucas Tree Model. 1

ASSET PRICING, INTEREST RATES, EXPECTED INFLATION, AND MONETARY POLICY 2

utility maximizing plan. We will assume that the period utility function u () satises all the usual properties (i.e. monotonicity, dierentiability, concavity, Inada, etc.). Note that the return from the purchase of a tree is given by pt+1 + dt+1 1 + rt+1 , pt which from the perspective of date t is uncertain, since dt+1 is not known at that time. Therefore, the Euler equation characterizing the agents optimal intertemporal consumption behavior is given by pt+1 + dt+1 u (ct ) = Et u (ct+1 ) . pt This expression can be re-written as pt = Et [Qt+1 (pt+1 + dt+1 )] , where Qt+1 u (ct+1 ) u (ct )

is the agents marginal rate of substitution between t and t + 1. As usual, we will be working with a representative agent economy. Therefore, in equilibrium asset prices must be such that the asset market clears: at = 1 for all t. Which in turn implies that ct = dt for all t. Therefore, in equilibrium u (dt+1 ) pt = E t (pt+1 + dt+1 ) u (dt ) u (dt+1 ) u (dt+1 ) = Et dt+1 + Et pt+1 . u (dt ) u (dt ) Since pt+1 will have a similar expression to that of pt , we can use repeated substitution to nd that u (dt+j ) u (dt+T ) pt = E t j dt+j + lim Et T pt+T . T u (dt ) u (dt ) j=1 "Bubble"
"Fundamental Value"

That is, the price of a tree at time t is equal to its fundamental value, the expected present discounted value of dividends, plus a bubble term. Q: Can we establish that the bubble term is zero? Two lines of argument may be used, within this model, to establish that the bubble term is zero: (1) Transversality. (2) Market clearing.

ASSET PRICING, INTEREST RATES, EXPECTED INFLATION, AND MONETARY POLICY 3

The rst is a purely technical argument, while the second relies on the underlying economics of the model. Note that the solution to the agents utility maximization problem is not only characterized by the Euler equation, but also by the transversality condition. That is, the Euler equation depcits the optimal intertemporal trade-o made by the agent, but the transversality condition is a restriction in the maximization problem, that is, it is a condition which restricts the type of consumption plans that the agent can choose. In this simple model the transversality condition is given by (T V C) : lim Et T u (ct+T ) pt+T = 0.
T

Using the market clearing condition ct = dt , and dividing by the constant u (dt ), we obtain T u (dt+T ) lim Et pt+T = 0. T u (dt ) Thus, we have shown that transversality implies that the bubble term is equal to zero. Now, for our second argument, remember that in equilibrium the price of trees must be consistent with market clearing. That is, at the prevailing price the representative agent must be willing to hold his tree. This means that we may write u (dt ) pt = Et j u (dt+j ) dt+j + lim Et T u (dt+T ) pt+T ,
j=1 T

where u (dt ) pt is the marginal benet of selling the tree, and Et is the marginal cost to the agent of retaining the tree forever.

j=1

j u (dt+j ) dt+j

Suppose that the bubble term was positive. This would imply that the marginal benet of selling the tree today exceeds its marginal cost. Therefore, all agents would want to sell their tree, exerting downward pressure on its price. On the otherhand, if the bubble term where negative, then the marginal benet of selling the tree would be smaller than its marginal cost and all agents would want to buy trees, thus exerting upward pressure on the price of trees. Therefore, the only scenario in which all agents want to hold on to the tree they own is the case in which the bubble term is exactly equal to zero. Therefore, we have established that in equilibrium the price of a tree is given by pt = E t St+j dt+j ,
j=1

where St+k =

j=1

Qt+j

ASSET PRICING, INTEREST RATES, EXPECTED INFLATION, AND MONETARY POLICY 4

is known in the nance literature as the stochastic discount factor, which is the way in which the agent brings future payos to present value. Notice that in this consumption-based asset pricing model we have that u (ct+k ) St+k = k , u (ct ) which is the agents marginal rate of substitution between t and t + k. We conclude that the equilibrium price of an asset is equal to its fundamental value.2 Example. Log Utility Suppose that u (c) = ln (c). In this case, the stochastic discount factor will be given by ct Qt+1 , ct+1 so that pt =

Et

dt Et j dt+j dt+j j=1 Et j dt = dt . 1 j=1


j=1

Qt+j dt+j

Notice that in this case, the state of the economy today (dt ) is mapped into the price of trees today (pt ), in a time invariant manner. Example. Mehra and Prescott (1985) Suppose that u (c) = c1 / (1 ) and dt+1 = t+1 dt , where evolves according to a Markov Chain with transition matrix P . In this case dt+1 pt = E t (pt+1 + dt+1 ) , dt which can also be written as pt pt+1 1 = Et (t+1 ) +1 dt dt+1 or, alternatively, as i =
j

Pij 1 (1 + j ) , j

2This of course, need not be true in other models of asset pricing. In some models, it is possible for the bubble term to be dierent from zero so that the price of an asset diverges from its fundamental value. See Santos and Woodford [1997] for cases in which asset bubbles may arise.

ASSET PRICING, INTEREST RATES, EXPECTED INFLATION, AND MONETARY POLICY 5

where is the price-to-dividend ratio, and Pij is the probability of transitioning from state i into state j. This expression was used by Mehra and Prescott [1985] to study the behavior of asset prices in the U.S. The Market Price of Risk. In a famous article, Hansen and Jagannathan [1991] studied certain implications of the Lucas asset pricing model to understand the market price of risk. To derive the result of interest we will make use of the concept of covariance and of a result known as the Cauchy-Schwarz Inequality. The covariance between two variables, say X and Y , is a measure of how much these two variables change together. The covariance between X and Y is dened as Cov (X, Y ) = E [(X E [X]) (Y E [Y ])] , and it is easy to show that Cov (X, Y ) = E [XY ] E [X] E [Y ]. The Cauchy-Schwarz Inequality states that for two variables X and Y , it is true that |E [XY ]| (E [X 2 ] E [Y 2 ]). Proof. Notice that for any R, we have that 2 0 E (X Y ) = E X 2 2E [XY ] + 2 E Y 2 . Let E X 2 /E [XY ]. Then the above expression becomes 2 2 2 E X2 0 E X 2E X + E Y2 , E [XY ] which implies E X
2

Taking square root on both sides of this last expression yields the desired result.

E2 X 2 E Y 2 E2 [XY ] E X 2 E Y 2 2 [XY ] E

Now, let pt be the time t price of an asset that delivers a random payo xt+1 at t + 1. Using the agents Euler equation we have that pt = Et [Qt+1 xt+1 ] , and from the denition of covariance we have Et [Qt+1 xt+1 ] = Covt (Qt+1 , xt+1 ) + Et [Qt+1 ] Et [xt+1 ] . Additionally, the Cauchy-Schwarz inequality implies |Covt (Qt+1 , xt+1 )| 1, t (Qt+1 ) t (xt+1 ) where denotes standard deviation.

ASSET PRICING, INTEREST RATES, EXPECTED INFLATION, AND MONETARY POLICY 6

Putting these three results together we may obtain the following expression pt t (Qt+1 ) Et [xt+1 ] t (xt+1 ) , Et [Qt+1 ] pf t where pf = Et [Qt+1 ] is the price of an asset whose payo is equal to 1 with t certainty. Therefore, pt /pf is the price of a risky asset relative to a risk-free asset. t In the nance literature3 this inequality is commonly written as f Et [Rt+1 ] Rt+1 t (Qt+1 ) , t (Rt+1 ) Et [Qt+1 ]
"Sharpe Ratio" Market price of risk

Figure 1 illustrates these bounds.

To understand why the term tt(Qt+1 ) is called the market price of risk, consider E [Qt+1 ] the case in which the inequality is actually an equality. In that case, pt /pf t t (Qt+1 ) = , t (xt+1 ) Et [Qt+1 ] so that
t (Qt+1 ) Et [Qt+1 ]

tells us how relative prices must change when the riskiness of the

payo of the risky asset increases.4 That is, tt(Qt+1 ) is the compensation that E [Qt+1 ] the market needs to oer investors in order for them to bear additional risk. Interest Rates and Expected Inflation For the eective functioning of monetary policy, it is important for the central bank to know (and sometimes aect) the publics ination expectations. The central banks actions respond to the publics ination expectations, and in an ination targeting scheme an important job for the central bank is to anchor expectations. But, how does the central bank elicit this information? From which economic indicators can it discern peoples ination expectations? Typically, movements in long-term bond yields are taken as reecting movements in the underlying ination expectations of the public. Since expectations aect behaviour, ination expectations will be embedded in the choices that economic agents make. In this section
3In the nance literature, the Sharpe ratio provides a measure of the excess returns, relative to volatility, that can be expected by investors if they hold the risky asset. 4The market price of risk is equal to minus the coecient of variation of the stochastic discount factor. If X is a random variable, then the coecient of variation of X is denes as Var (X) Std (X) CV (X) = = . E [X] E [X]

The coecient of variation provides a normalized measure of dispersion for a random variable (notice that this measure of dispersion really only makes sense for non-negative random variables. Otherwise we risk division by zero or obtaining a negative number as the measure of dispersion).

ASSET PRICING, INTEREST RATES, EXPECTED INFLATION, AND MONETARY POLICY 7

Figure 0.1. Hansen-Jagannathan Bounds we will analyze the relationship between nominal interest rates and ination expectations, and we will try to use theory to decompose movements in long-term bond yields into changes in ination expectations and other determinants. The Fisherian Theory of Interest Rates. The Fisherian theory of interest rates can be very simply, and intuitively, derived from the optimal allocation of resources by investors. Consider an investor who can allocate his resources to either a nominal bond (one that pays in dollar terms) or to a real bond (one that pays in units of the cosumption good). The investment technology faced by the investor is given by Asset Nominal Bond Real Bond t = 0 Price $1 1 unit of consumption good t = 1 Return $1 + R 1 + r units of consumption good

Attached to the nominal bond is a nominal interest rate equal to R, and attached to the real bond is a real interest rate equal to r. The question is, how does the investor want to allocate his wealth across these two assets? Clearly, the investor will wish to invest in the asset with the highest rate of return. However, notice that one asset pays in dollars and the other pays in units of the consumption good, these units are not comparable so we must rst express the return of both assets in terms of a common unit of measurement (either dollars or consumption goods). If at date t = 0 the price level is P0 , then the investor can sell 1/P0 units of the consumption good to obtain 1 dollar, which he can use to buy one unit of the nominal bond. Therefore, the date 0 price of the nominal bond, in terms of consumption goods, is 1/P0 . At date t = 1, the investor will receive R dollars for each nominal bond he holds. The value of this payo, in terms of consumption goods, depends on the price level that prevails at t = 1. If the investor expects that the date t = 1 price e level will be P1 , then his date t = 1 payo, in terms of consumption goods, will be

ASSET PRICING, INTEREST RATES, EXPECTED INFLATION, AND MONETARY POLICY 8 e 1 + R/P1 . This means that the real rate of return of a nominal bond is given by 1+R e P0 1+R P1 = (1 + R) = , 1 e P1 1 + e P0

where e is the ination expected by the investor between t = 0 and t = 1. If 1+R > 1 + r, 1 + e

then all investors would wish to invest all their wealth in the nominal bond. If the opposite inequality holds, then all investors would wish to allocate all their resources to the real bond. Neither of these can be equilibrium situations.5Equilibrium requires the no arbitrage condition 1+R = 1 + r, 1 + e that is, prices adjust so that the rate of return of both assets is equal. The no arbitrage condition can also be written as R r + e , by taking logs and using the approximation ln (1 + x) x. From this expression we can see that R r + e . If r 0 (i.e. the real interest rate is fairly stable), the changes in the nominal interest rate can be taken as reecting mostly changes in ination expectations. Going a step further, from a two period consumption-savings model we know that if the agents utility function over date t = 0 and date t = 1 consumption is of the form 1 c1 1 1 c1 1 U (c0 , c1 ) = 0 + , 1 1+ 1 then the Euler equation, which characterizes the optimal intertemporal consumption decision of the agent, can be written as r = [ln (c1 ) ln (c0 )] + . From this expression we easily see that the real interest rate will be fairly stable whenever the growth rate of consumption is stable, and that even when the real interest rate is not directly observable, it can be inferred from the growth rate of consumption (provided we know and ). The Fisherian theory of interest rates is simple and intuitive, yet it omits a key element which aects investors decisions: risk. We will now use the Lucas asset pricing model to further our understanding of the determinants of nominal interest rates. To this end, consider a version of the Lucas model in which there are two assets: nominal bonds Bt , each unit of which pays 1 + Rt+1 dollars at t + 1 with
5Why?

ASSET PRICING, INTEREST RATES, EXPECTED INFLATION, AND MONETARY POLICY 9

certainty, and real bonds bt which pay 1 + rt+1 units of the consumption good at t + 1. In this case, the budget constraint faced by the representative agent is given by Bt (1 + Rt1 ) Bt1 c t + bt + yt + (1 + rt1 ) bt1 + . Pt Pt Here, the only uncertainty faced by the consumer is that at time t he does not know the price level that will prevail at t + 1, which makes the real payo of the nominal bond random. In this case the optimal investment decisions are captured by two Euler equations: 1 = Et [Qt+1 ] 1 + rt 1 Pt = Et Qt+1 , 1 + Rt Pt+1 where recall that Qt+1 is the stochastic discount factor. Using the denition of covariance we may derive the following expression: 1 1 1 1 = Et + Covt Qt+1 , 1+R 1+r 1 + t+1 1 + t+1 t t
Nominal Rate Real Rate Expected Ination Risk

It is readily seen that the above expression generalizes the Fisherian theory of interest rates to account for the risk that investors have to bear. This terms appears as a determinant of the nominal interest rate since random movements in ination make the goods-denominated return of a nominal bond uncertain. To gain further intuition, suppose u (c) = ln (c). In this case 1 ct 1 Covt Qt+1 , = Covt , , 1 + t+1 ct+1 1 + t+1 which can be positive or negative depending on how consumption growth covaries with ination. For example, if Covt (ct+1 , t+1 ) < 0 so that periods of high ination correspond to periods where the agents consumption is growing slowly, then the risk term is negative meaning that the nominal interest rate would be higher than that implied by the Fisherian theory, precisely because investors are being compensated for the risk that they have to bear, which in this case aects them more severely at those times when they need income the most. With this generalization of the determination of nominal interest rates, it can be seen that movements in long-term bond yields accurately reect changes in expected ination if and only if the real interest rate is stable and the risk premium is small. Using the same arguments that we used to derive the implications of the Lucas model for the market price of risk, we can establish bounds on expected ination. The Cauchy-Schwarz inequality, together with the assumption that ination volatility is bounded in the sense that t (1/1 + t+1 ) Et [1/1 + t+1 ], implies that ct 1 ct 1 ct 1 Et Covt , t Et . t ct+1 1 + t+1 ct+1 1 + t+1 ct+1 1 + t+1

ASSET PRICING, INTEREST RATES, EXPECTED INFLATION, AND MONETARY POLICY 10

Recall that Covt

1 , ct+1 1 + t+1

ct

1 ct 1 = Et Et , 1 + Rt ct+1 1 + t+1

so that we may arrive at ct ct 1 ct ct (1 + Rt ) Et (1 + Rt ) Et t + t . 1 ct+1 ct+1 ct+1 ct+1 Et 1+t+1 Locally we may use the approximation 1 Et [1 + t+1 ] , 1 Et 1+t+1

so that we may establish the following bounds on expected ination: ct ct ct ct (1 + Rt ) Et t Et [1 + t+1 ] (1 + Rt ) Et + t . ct+1 ct+1 ct+1 ct+1 Observe that, given , these bounds may be calculated from data on nominal interest rates and aggregate consumption. Asset Prices, Asset Price Bubbles and Monetary Policy In recent decades there have been a number of episodes of rapid asset price ination often followed by a nancial and economic crisis. In fact, the remarkable rise in real-estate prices in many advanced economies and its ensuing collapse are generally viewed as the key detonating factor underlying the global nancial crisis of 20072009. It is for these reasons that recent debates have focused on the appropriate response of monetary policy to developments in asset markets, as low and stable ination has not proven to guarantee nancial stability. The consensus amongst policy makers was that the monetary authority should respond to asset prices only insofar as they conveyed information about the future path of output and ination which are the goal variables of monetary policy. However, recent experience has called into question the wisdom of this approach, and it is now argued that the central bank should act pre-emptively in the face of developments in asset markets. Proponents of this bubble policy approach to the conduct of monetary policy in response to asset prices argue that movements in the bubble component of asset prices can have serious adverse consequences for macroeconomic performance that monetary policy may not be able to readily oset after the fact. Therefore, it is advised that the monetary authority should eliminate this source of risk directly. Under the bubble policy approach, a constraint faced by the monetary authority in the pursuit of its goals of low and stable ination and economic growth is the nancial enviornment in which it operates, and it is possible for them to take actions that aect this enviornment in order that they may more successfully achieve their goals. Interventions in asset markets would thus entail a trade-o for the monetary

ASSET PRICING, INTEREST RATES, EXPECTED INFLATION, AND MONETARY POLICY 11

authority in which they would incur in short-term costs in terms of deviations from the central banks macroeconomic goals, in favor of long-term benets in terms of reducing the future risk of a nancial crisis whose eects would be dicult to correct after the fact, and that would hinder the future ability of the monetary authority to achieve its objectives. There is a number of questions that policymakers must consider before they decide to respond to developments in asset prices: (1) Can policymakers identify bubbles in asset prices? (2) Does the monetary authority have the ability to intervene in asset markets to contain asset price bubbles? (3) Will fallout from a bubble, which may include nancial crisis and/or macroeconomic imbalances and resource misallocation, be signicant and hard to rectify after the fact? (4) If an asset price bubble has been identied in the price of a particular asset, can policy makers be sure that interventions by the monetary authority designed to contain this bubble will not adversely aect the prices for other assets which are not aected by a bubble? (5) Is monetary policy the best tool to deate the bubble? When it has been agreed that the monetary authority should intervene to contain a bubble in asset prices, the traditional view recommends higher interest rates to oset any economic stimulus generated by the bubble, and to dampen or bring to an end any episodes of speculative frenzy that may have caused the bubble in the rst place. However, in light of the renewed interest in the design of optimal policy responses to asset price bubbles, this conventional view has been strongly questioned.6 To gain some intuition behind the critique of the traditonal view of policy responses to asset price bubbles, consider the following simplied partial equilibrium framework: there is a representative risk-neutral investor (i.e. his preferences are of the type u (c) = c, for > 0) who can borrow and lend at the riskless rate Rt . Let pt be the time t price of an innitely-lived asset with dividend stream {dt+i }i=0 . Recall from our discussion of the Lucas asset pricing model that the price of this asset must satisfy pt = Et [Qt+1 (pt+1 + dt+1 )] , where Qt+1 is the stochastic discount factor, equal to the agents marginal rate of 1 substitution. Here, since utility is linear, we have that Qt+1 = (1 + Rt ) (i.e. the
6One of the reasons why optimal policy responses to asset price bubbles are so poorly un-

derstood, from the theoretical point of view, is that the dominant paradigm in macroeconomic modelling is the innitely lived representative agent framework, and within this framework rational asset price bubbles can typically be ruled out under standard assumptions (see Santos and Woodford, 1997). To allow for the possibility of rational asset price bubbles to exist in equilibrium, one must work within the overlapping generations framework which is typically only used when addressing issues concerned with social security.

ASSET PRICING, INTEREST RATES, EXPECTED INFLATION, AND MONETARY POLICY 12

investor discounts between today and tomorrow based only on the riskless rate R). Thus, we have that the asset price must satisfy the condition pt (1 + Rt ) = Et [pt+1 + dt+1 ] . Now, suppose that the asset price may be decomposed into its fundamental component and its bubble component: pt = p F + p B . t t Again, based on our discussion of the Lucas asset pricing model we know that the fundamental component of an assets price is the expected discounted present value of dividends: F pt = E t St+k dt+k ,
k=1

where St+k =

k1 j=0

1 (1 + Rt+j )

is the price of time t + k goods in terms of time t goods. Therefore, we may write the equilibrium condition that asset prices must satisfy as pF (1 + Rt ) + pB (1 + Rt ) = Et dt+1 + pF + Et pB . t t t+1 t+1 By using the expression above for the fundamental component of asset prices, it may be shown that pF (1 + Rt ) = Et dt+1 + pF , t t+1 so that in equilibrium the bubble component of asset prices must satisfy pB (1 + Rt ) = Et pB , t t+1 or stated dierently Et pB t+1 = (1 + Rt ) . pB t

This last condition states that in equilibrium the expected growth rate of the bubble component of asset prices is controlled by the interest rate. Therefore, ceteris paribus, an increase in the interest rate will lower pF , the fundamental component t of asset prices, since dividends will be discounted more heavily, while this increase in the interest rate will allow the bubble component to grow more rapidly (in expectation). Therefore, policies that imply a systematic positive response of the interest rate to the size of the bubble will tend to amplify the movements in the latter. This relation between monetary policy and asset price bubbles is clearly at odds with the traditional view, which recommends higher interest rates as the natural way to disinate a growing bubble. Gal (2011) develops these ideas in a fully specied dynamic general equilibrium model in which there are nominal rigidities (so monetary policy may aect the real

ASSET PRICING, INTEREST RATES, EXPECTED INFLATION, AND MONETARY POLICY 13

interest rate) and nominal interest rate policy is conducted through a policy rule that determines the systematic response of nominal interest rates to movements in the size of the asset price bubble. In such a framework, Gal shows that: Monetary policy cannot aect the conditions for existence of an asset price bubble, but it can inuence its short-run behavior, and in particular the size of its uctuations. Contrary to the conventional wisdom, a stronger interest rate response to bubble uctuations may raise the volatility of asset prices and of their bubble component. The optimal policy must reach a balance between stabilization of current aggregate demand (which calls for a positive interest rate response to the bubble) and stabilization of the bubble itself (which would dictate a negative interest rate response to the bubble). References
[1] Alvarez, Fernando and Urban J. Jermann. (2004). Using Asset Prices to Measure the Costs of Business Cycles. Journal of Political Economy. [2] Cochrane, John H. and Lars Peter Hansen. (1992). Asset Pricing Explorations for Macroeconomics. NBER Macroeconomics Annual vol.7. [3] Gal, Jordi. (2011). Monetary Policy and Rational Asset Price Bubbles. Working Paper, Department of Economics, Universitat Pompeu Fabra. [4] Ireland, Peter N. (1996). Long-Term Interest Rates and Ination: A Fisherian Approach. Federal Reserve Bank of Richmond Economic Quarterly, Volume 82. [5] Hansen, Lars P. and Ravi Jagannathan. (1991). Implications of Security Market Data For Models of Dynamic Economics. Journal of Political Economy. [6] Lucas, Robert E. Jr. (1978). Asset Prices in an Exchange Economy. Econometrica. [7] Santos, Manuel S. y Michael Woodford. (1997). Rational Asset Pricing Bubbles. Econometrica. [8] Stock, James H. and Mark W. Watson. (2003). Forecasting Output and Ination: The Role of Asset Prices. Journal of Economic Literature. [9] Vickers, John. (1999). Monetary Policy and Asset Prices. Money, Macro and Finance Group 31st Annual Conference.

You might also like