Samuelson's model (1939) combines the multiplier and the accelerator to show how a once-for-all disturbance can generate a self-sustained or damped business cycle. Income in each period is the sum of consumption, induced investment and autonomous (government) spending.

Assumptions

• Consumption depends on last period's income: $C_t = a + bY_{t-1}$, with $0<b<1$ ($b$ = MPC).
• Induced investment depends on the change in consumption (the accelerator): $I_t = v(C_t - C_{t-1}) = vb(Y_{t-1}-Y_{t-2})$, with $v>0$ the accelerator coefficient. (Writing $\beta = vb$ gives the usual textbook form.)
• Autonomous spending $G_t = G_0$ is constant.

Derivation of the income difference equation

$Y_t = C_t + I_t + G_0 = a + bY_{t-1} + vb\,(Y_{t-1}-Y_{t-2}) + G_0$

$\boxed{Y_t - b(1+v)\,Y_{t-1} + bv\,Y_{t-2} = a + G_0}$

Equilibrium (set $Y_t=Y_{t-1}=Y_{t-2}=Y^*$):

$Y^* = \frac{a+G_0}{1-b}$

which is the simple multiplier solution; the accelerator does not change the equilibrium level, only the path towards it.

Roles of MPC and accelerator

• The MPC ($b$) drives the multiplier: a rise in income raises consumption, which raises income again in the next period.
• The accelerator ($v$) makes investment respond to the change in demand. A rising income raises investment sharply, which amplifies the boom; a slowdown cuts investment, which deepens the slump.
• Multiplier alone gives monotonic adjustment; only the interaction of the two can give cycles.

Solution and conditions

The characteristic equation is

$\lambda^2 - b(1+v)\lambda + bv = 0,\qquad \lambda_{1,2}=\frac{b(1+v)\pm\sqrt{b^2(1+v)^2-4bv}}{2}$

with $Y_t = Y^* + A_1\lambda_1^t + A_2\lambda_2^t$.

• Stability: both roots must lie inside the unit circle. For this equation the conditions reduce to $bv<1$ (the others, $1-b(1+v)+bv=1-b>0$ and $1+b(1+v)+bv>0$, always hold). Time path is convergent if $bv<1$, uniform if $bv=1$, explosive if $bv>1$.
• Cycles: the roots are complex when the discriminant is negative, $b^2(1+v)^2<4bv$, i.e. $b(1+v)^2<4v$. Then the path oscillates with modulus $\sqrt{bv}$.

Five cases (with $b$, $v$ plotted in the plane)

The boundary between cyclical and non-cyclical regions is the curve $b(1+v)^2=4v$.

Numerical illustration: $b=0.5,\ v=1$ gives $\lambda^2 - \lambda + 0.5=0$, so $\lambda = 0.5\pm 0.5i$, modulus $\sqrt{0.5}\approx0.71<1$: damped cycles around $Y^*$.

Conclusion: A high MPC and a high accelerator together make the economy cyclical and potentially explosive; stability requires $bv<1$.

Setup. The consumer maximises $U=f(x,y)$ subject to $B=p_xx+p_yy$. The first-order conditions $f_x=\lambda p_x,\ f_y=\lambda p_y$ and the budget constraint give the Marshallian demand $x=x(p_x,p_y,B)$.

Derivation. Let $h_x(p_x,p_y,\bar U)$ be the Hicksian (compensated) demand, the demand that minimises expenditure for a given utility $\bar U$, with expenditure function $e(p_x,p_y,\bar U)$. At the optimum, $B=e(p_x,p_y,\bar U)$ and

$h_x(p_x,p_y,\bar U)=x\big(p_x,p_y,e(p_x,p_y,\bar U)\big).$

Differentiate with respect to $p_x$, holding $p_y$ and $\bar U$ fixed:

$\frac{\partial h_x}{\partial p_x}=\frac{\partial x}{\partial p_x}+\frac{\partial x}{\partial B}\cdot\frac{\partial e}{\partial p_x}.$

By Shephard's lemma, $\partial e/\partial p_x=h_x=x$. Rearranging gives the Slutsky equation

$\boxed{\frac{\partial x}{\partial p_x}=\underbrace{\frac{\partial h_x}{\partial p_x}}_{\text{substitution effect}}-\underbrace{x\,\frac{\partial x}{\partial B}}_{\text{income effect}}}$

The substitution effect is always $\le 0$ (compensated demand slopes downward). The income effect is the loss of real income caused by the price rise, $x\,dp_x$, multiplied by the income response $\partial x/\partial B$.

Elasticity form. Multiply by $p_x/x$ and write the last term as $\dfrac{p_xx}{B}\cdot\dfrac{\partial x}{\partial B}\dfrac{B}{x}$:

$e_{xx}=e^{s}_{xx}-s_x\,e_{xB}$

where $e_{xx}$ is the own-price elasticity, $e^{s}_{xx}=\dfrac{\partial h_x}{\partial p_x}\dfrac{p_x}{x}\le0$ the compensated (substitution) elasticity, $e_{xB}$ the income elasticity and $s_x=p_xx/B$ the budget share.

Signs of the effects (for a price rise)

For a Giffen good, $e_{xx}>0$ requires $-s_xe_{xB}>\lvert e^{s}_{xx}\rvert$, so the good must have a large budget share and a strongly negative income elasticity. Staple foods for very poor households are the usual example.

Reading the data as $A=\begin{pmatrix}3/10&4/10\\3/10&2/10\end{pmatrix}$ and $d=\begin{pmatrix}12/10\\12/10\end{pmatrix}$, the continuous-time output adjustment model is

$\dot x = Ax+d-x=-(I-A)x+d,\qquad I-A=\begin{pmatrix}0.7&-0.4\\-0.3&0.8\end{pmatrix},\ |I-A|=0.44.$

(a) Particular integral (the equilibrium $\bar x$ with $\dot x=0$):

$\bar x=(I-A)^{-1}d=\frac{1}{0.44}\begin{pmatrix}0.8&0.4\\0.3&0.7\end{pmatrix}\begin{pmatrix}1.2\\1.2\end{pmatrix}=\frac1{0.44}\begin{pmatrix}1.44\\1.20\end{pmatrix}=\begin{pmatrix}36/11\\30/11\end{pmatrix}\approx\begin{pmatrix}3.27\\2.73\end{pmatrix}$

(b) Complementary function. Solve $\dot x=(A-I)x$ with $A-I=\begin{pmatrix}-0.7&0.4\\0.3&-0.8\end{pmatrix}$.

Characteristic equation: $\lambda^2+1.5\lambda+0.44=0\Rightarrow(\lambda+0.4)(\lambda+1.1)=0$, so $\lambda_1=-0.4,\ \lambda_2=-1.1$.

Eigenvectors: for $\lambda_1=-0.4$, $-0.3v_1+0.4v_2=0\Rightarrow v=(4,3)'$; for $\lambda_2=-1.1$, $0.4v_1+0.4v_2=0\Rightarrow v=(1,-1)'$.

$x_c=c_1\begin{pmatrix}4\\3\end{pmatrix}e^{-0.4t}+c_2\begin{pmatrix}1\\-1\end{pmatrix}e^{-1.1t}$

(c) Definite time path. The general solution is $x(t)=\bar x+x_c$. At $t=0$:

$4c_1+c_2=\frac{187}{38}-\frac{36}{11}=\frac{689}{418},\qquad 3c_1-c_2=\frac{72}{13}-\frac{30}{11}=\frac{402}{143}$

Adding: $7c_1=\dfrac{689}{418}+\dfrac{402}{143}$, so $c_1\approx0.637$ and $c_2\approx-0.900$.

$x_1(t)\approx3.27+2.55\,e^{-0.4t}-0.90\,e^{-1.1t},\qquad x_2(t)\approx2.73+1.91\,e^{-0.4t}+0.90\,e^{-1.1t}$

Both roots are negative, so the system is stable and converges monotonically to $\bar x=(36/11,\,30/11)'$, with the slower root $-0.4$ dominating in the long run.

Shephard's Lemma

Statement: If the expenditure function is $e(p, u)$, then the Hicksian (compensated) demand for good $i$ is obtained simply by partially differentiating the expenditure function with respect to the price of that good:

$x_i^h(p, u) = \frac{\partial e(p, u)}{\partial p_i}$

Derivation:

Let the consumer minimise expenditure subject to attaining a fixed utility level $u$:

$\min_{x} \; p \cdot x \quad \text{subject to} \quad u(x) = u$

The expenditure function is the value of this minimisation problem:

$e(p, u) = \min_{x} \{ p \cdot x : u(x) = u \} = p \cdot x^h(p, u)$

where $x^h(p, u)$ is the vector of Hicksian demand functions.

Now differentiate $e(p, u)$ with respect to $p_i$ using the chain rule. Since $e(p, u) = \sum_j p_j x_j^h(p, u)$:

$\frac{\partial e(p, u)}{\partial p_i} = x_i^h(p, u) + \sum_j p_j \frac{\partial x_j^h(p, u)}{\partial p_i}$

The second term vanishes by the envelope theorem. At the optimum, the first-order conditions of the expenditure-minimisation problem are:

$p_j = \lambda \frac{\partial u}{\partial x_j} \quad \text{for all } j$

where $\lambda$ is the Lagrange multiplier on the utility constraint. Substituting these into the second term:

$\sum_j p_j \frac{\partial x_j^h}{\partial p_i} = \lambda \sum_j \frac{\partial u}{\partial x_j} \cdot \frac{\partial x_j^h}{\partial p_i} = \lambda \frac{d}{dp_i}\big[ u(x^h(p, u)) \big]$

But along the Hicksian demand curve utility is held constant at $u$, so $u(x^h(p, u)) = u$ for all $p$. Hence its derivative with respect to $p_i$ is zero:

$\lambda \cdot 0 = 0$

Therefore the second term vanishes and we are left with:

$\boxed{\frac{\partial e(p, u)}{\partial p_i} = x_i^h(p, u)}$

Interpretation: The partial derivative of the expenditure function with respect to a price gives the compensated demand for that good. It tells us how much extra money the consumer needs to maintain the same utility level when the price of good $i$ rises by one unit — which is exactly the quantity of good $i$ she is buying. This is why the result is also called the "compensated demand = price derivative of expenditure" property.

Cobb–Douglas production function

The Cobb–Douglas production function is a special form of production function that shows the technological relationship between inputs and output. For two inputs, labour $(L)$ and capital $(K)$, it is written as:

$$Q = A L^{\alpha} K^{\beta}$$

where:

• $Q$ = total output,
• $A$ = efficiency/technology parameter $(A>0)$,
• $\alpha$ = output elasticity of labour,
• $\beta$ = output elasticity of capital,
• $\alpha,\beta > 0$.

It is homogeneous of degree $\alpha+\beta$. Thus:

• if $\alpha+\beta=1$, constant returns to scale,
• if $\alpha+\beta>1$, increasing returns to scale,
• if $\alpha+\beta<1$, decreasing returns to scale.

---

1. Marginal productivities (MP)

Marginal product of labour $(MP_L)$:

$$MP_L = \frac{\partial Q}{\partial L} = \alpha A L^{\alpha-1} K^{\beta} = \alpha \frac{Q}{L} = \alpha AP_L$$

Marginal product of capital $(MP_K)$:

$$MP_K = \frac{\partial Q}{\partial K} = \beta A L^{\alpha} K^{\beta-1} = \beta \frac{Q}{K} = \beta AP_K$$

---

2. Productivity function (average productivity)

Average productivity of labour:

$$AP_L = \frac{Q}{L} = A L^{\alpha-1} K^{\beta}$$

Average productivity of capital:

$$AP_K = \frac{Q}{K} = A L^{\alpha} K^{\beta-1}$$

---

3. Output elasticity

$$E_L = \frac{\partial Q}{\partial L}\cdot \frac{L}{Q} = \alpha$$ $$E_K = \frac{\partial Q}{\partial K}\cdot \frac{K}{Q} = \beta$$

So the sum of output elasticities is:

$$E_L + E_K = \alpha+\beta$$

which also gives the degree of homogeneity and hence the returns to scale.

Conclusion: For $Q = A L^{\alpha} K^{\beta}$, the marginal products are $MP_L=\alpha Q/L$ and $MP_K=\beta Q/K$, while the average productivity functions are $AP_L=Q/L$ and $AP_K=Q/K$.

Let
[ u_t=\begin{pmatrix}x_t\ y_t\end{pmatrix},\qquad A=\begin{pmatrix}3&-2\2&-1\end{pmatrix},\qquad c=\begin{pmatrix}12\6\end{pmatrix}. ] Then the system is [ u_{t+1}=Au_t+c. ]

The characteristic equation of (A) is [ |A-\lambda I|= \begin{vmatrix} 3-\lambda&-2\ 2&-1-\lambda \end{vmatrix} =(3-\lambda)(-1-\lambda)+4 =\lambda^2-2\lambda+1=(\lambda-1)^2. ] So (\lambda=1) is a repeated eigenvalue.

Now [ N=A-I= \begin{pmatrix} 2&-2\ 2&-2 \end{pmatrix}, \qquad N^2=0. ] Hence [ A^t=(I+N)^t=I+tN= \begin{pmatrix} 1+2t&-2t\ 2t&1-2t \end{pmatrix}. ]

The general solution is [ u_t=A^t u_0+\sum_{j=0}^{t-1}A^j c. ]

First, [ A^t u_0= \begin{pmatrix} 1+2t&-2t\ 2t&1-2t \end{pmatrix} \begin{pmatrix} 4\3 \end{pmatrix}

\begin{pmatrix} 4+2t\ 3+2t \end{pmatrix}. ]

Also, [ \sum_{j=0}^{t-1}A^j c

\left(tI+\frac{t(t-1)}2N\right)c. ] Since [ Nc= \begin{pmatrix} 2&-2\ 2&-2 \end{pmatrix} \begin{pmatrix} 12\6 \end{pmatrix}

\begin{pmatrix} 12\12 \end{pmatrix}, ] we get [ \sum_{j=0}^{t-1}A^j c

t\begin{pmatrix}12\6\end{pmatrix} + \frac{t(t-1)}2\begin{pmatrix}12\12\end{pmatrix}

\begin{pmatrix} 6t^2+6t\ 6t^2 \end{pmatrix}. ]

Therefore, [ u_t= \begin{pmatrix} 4+2t\ 3+2t \end{pmatrix} + \begin{pmatrix} 6t^2+6t\ 6t^2 \end{pmatrix}. ]

So the required solution is [ \boxed{x_t=6t^2+8t+4} ] [ \boxed{y_t=6t^2+2t+3} ]

Verification for (t=0): [ x_0=4,\qquad y_0=3, ] which satisfies the given initial conditions.

1. Setup. Let $Z=f(X,Y)$ be twice continuously differentiable, with first partials $f_X,f_Y$ and second partials $f_{XX},f_{YY},f_{XY}=f_{YX}$ (Young's theorem).

2. First-order (necessary) condition. At an interior extremum the total differential must vanish for arbitrary $dX,dY$:

$dZ=f_X\,dX+f_Y\,dY=0\ \Rightarrow\ f_X=0,\ f_Y=0.$

This locates stationary points but cannot tell a maximum from a minimum or a saddle.

3. Second-order condition and the Hessian. The second differential is the quadratic form

$d^2Z=f_{XX}(dX)^2+2f_{XY}\,dX\,dY+f_{YY}(dY)^2,$

with Hessian matrix and determinant

$H=\begin{pmatrix}f_{XX}&f_{XY}\\f_{YX}&f_{YY}\end{pmatrix},\qquad |H|=f_{XX}f_{YY}-(f_{XY})^2.$

4. Derivation of the sign conditions. Completing the square (for $f_{XX}\ne0$):

$d^2Z=\frac{1}{f_{XX}}\big(f_{XX}\,dX+f_{XY}\,dY\big)^2+\frac{f_{XX}f_{YY}-(f_{XY})^2}{f_{XX}}\,(dY)^2.$

The first term is a perfect square, so its sign follows the sign of $f_{XX}$. The second term's sign follows $|H|/f_{XX}$. Hence:

• Relative maximum: $d^2Z<0$ for all $(dX,dY)\ne0$ requires $f_{XX}<0$ and $|H|>0$.
• Relative minimum: $d^2Z>0$ requires $f_{XX}>0$ and $|H|>0$.
• Saddle point: $|H|<0$; the two terms have opposite signs, so $d^2Z$ changes sign with direction.
• Inconclusive: $|H|=0$; higher-order terms must be examined.

5. Summary

Equivalently, the leading principal minors alternate in sign for a maximum ($|H_1|<0,\ |H_2|>0$) and are all positive for a minimum ($|H_1|>0,\ |H_2|>0$). Note that $|H|>0$ forces $f_{XX}$ and $f_{YY}$ to have the same sign.

6. Interpretation. $|H|$ measures the curvature of the surface at the stationary point. If $|H|>0$ the surface curves the same way in every direction: dome-shaped (peak) when $f_{XX}<0$, bowl-shaped (trough) when $f_{XX}>0$. If $|H|<0$ it curves up in one direction and down in another, giving a saddle.

The model meant here is Domar's growth model (the Harrod–Domar framework). Its instability is the famous "knife-edge" result: the economy can grow in balance only at one particular rate, and any deviation from it is magnified, not corrected.

Assumptions: saving is a constant fraction of income, $S=sY$; the capacity-output relation is fixed, so each unit of investment raises potential output by $\sigma=1/v$ (the capital-output ratio $v$ is constant); no technical progress and no depreciation.

Demand side (multiplier). Equilibrium needs $I=S=sY$. If investment grows, income must grow by $\Delta Y=\dfrac{\Delta I}{s}$, so required income growth is

$\frac{\Delta Y}{Y}=\frac{\Delta I}{I}\cdot\frac{I}{sY}=\frac{\Delta I}{I}\quad(\text{since }I=sY).$

Supply side (capacity). Investment $I$ raises capacity by $\Delta K=I$, i.e. potential output increases by $\sigma I$.

Condition for balanced growth. Demand growth must equal capacity growth:

$\Delta Y=\sigma I\ \Rightarrow\ \frac{\Delta Y}{Y}=\sigma s=\frac{s}{v}.$

This is the warranted rate $G_w=s/v$. For example, with $s=0.20$ and $v=4$, $G_w=5\%$.

Why it is unstable (Harrod's argument). Suppose actual growth $G$ differs from $G_w$.

• If $G>G_w$: actual output grows faster than planned, so the actual capital-output ratio is below the desired one. Firms see capital as scarce and raise investment further, so $G$ rises even more above $G_w$. The result is cumulative inflationary boom.
• If $G<G_w$: capital is in excess, so firms cut investment, income growth falls further below $G_w$. The result is cumulative slump and unemployment.

Formally, the deviation $G-G_w$ grows over time (a positive feedback), so the equilibrium is unstable. The economy "grows on a razor's edge".

Second knife-edge. Even if $G=G_w$, steady full employment also needs $G_w=G_n=n$ (the natural rate). Since $s$, $v$ and $n$ are determined independently, equality is accidental; if $G_w>G_n$ there is secular stagnation, and if $G_w<G_n$ there is chronic unemployment.

Conclusion: because balanced growth requires $G=G_w=G_n$ and any departure is self-reinforcing, the Domar/Harrod–Domar model is highly unstable. This is why later models (Solow–Swan) allow $v$ to vary to restore stability.

Given: Maximize [ J=\int_0^1 (4x-3y^2),dt ] subject to [ \dot{x}=8y,\qquad x(0)=5,\qquad x(1)\ \text{free}. ]

Step 1: Form the Hamiltonian

For a maximization problem, [ H=4x-3y^2+\lambda(8y). ]

Step 2: Pontryagin necessary conditions

Control condition: [ \frac{\partial H}{\partial y}=-6y+8\lambda=0 ] [ \Rightarrow y=\frac{4}{3}\lambda. ]

Costate equation: [ \dot{\lambda}=-\frac{\partial H}{\partial x}=-4. ]

Transversality condition since (x(1)) is free: [ \lambda(1)=0. ]

Step 3: Solve the costate equation

[ \dot{\lambda}=-4,\qquad \lambda(1)=0. ] Integrating, [ \lambda(t)=-4t+C. ] Using (\lambda(1)=0), [ -4+C=0 \Rightarrow C=4. ] Thus, [ \lambda(t)=4(1-t). ]

Step 4: Find the optimal control

[ y(t)=\frac{4}{3}\lambda(t) =\frac{4}{3}\cdot 4(1-t) =\frac{16}{3}(1-t). ]

So, [ \boxed{y^*(t)=\frac{16}{3}(1-t)}. ]

Step 5: Find the state trajectory

[ \dot{x}=8y=\frac{128}{3}(1-t). ] Integrating, [ x(t)=x(0)+\int_0^t \frac{128}{3}(1-s),ds. ] Since (x(0)=5), [ x(t)=5+\frac{128}{3}\left(t-\frac{t^2}{2}\right). ] Therefore, [ \boxed{x^*(t)=5+\frac{128}{3}t-\frac{64}{3}t^2}. ]

At (t=1), [ x(1)=5+\frac{128}{3}\cdot \frac12 =5+\frac{64}{3} =\frac{79}{3}. ]

Step 6: Maximum value of the objective functional

[ J^=\int_0^1 \left(4x^-3y^{2}\right)dt. ] Substituting, [ 4x^-3y^{2} =20+\frac{512}{3}t-\frac{256}{3}t^2 -\frac{256}{3}(1-2t+t^2) ] [ =-\frac{196}{3}+\frac{1024}{3}t-\frac{512}{3}t^2. ] Thus, [ J^=\int_0^1\left(-\frac{196}{3}+\frac{1024}{3}t-\frac{512}{3}t^2\right)dt ] [ =-\frac{196}{3}+\frac{1024}{3}\cdot\frac12-\frac{512}{3}\cdot\frac13 ] [ =-\frac{196}{3}+\frac{512}{3}-\frac{512}{9} =\frac{436}{9}. ]

Hence, [ \boxed{J^*=\frac{436}{9}} ]

Final answer: [ \boxed{y^(t)=\frac{16}{3}(1-t)} ] [ \boxed{x^(t)=5+\frac{128}{3}t-\frac{64}{3}t^2} ] [ \boxed{J^*=\frac{436}{9}} ]

CGE Model (Computable General Equilibrium Model)

A CGE model is a numerical, economy-wide model that applies general equilibrium theory to actual data of an economy to compute the equilibrium prices and quantities of all markets simultaneously. It converts Walrasian general equilibrium into an empirically solvable system, using a Social Accounting Matrix (SAM) as its database. It is widely used to analyse the effects of policy shocks such as tax reform, trade liberalisation, subsidies, and public expenditure changes on output, prices, income distribution, and welfare.

Important components:

1. Social Accounting Matrix (SAM) — the data base that records all transactions among production activities, commodities, factors of production (labour, capital, land), households, government, and the rest of the world in a consistent double-entry format.

2. Production block — production functions (usually CES or Cobb–Douglas) describing how firms combine intermediate inputs and primary factors to produce output, along with the zero-profit condition (price = unit cost) under perfect competition.

3. Factor markets — supply and demand for factors of production, with factor prices determined endogenously; full employment or unemployment (e.g., efficiency-wage or fixed-wage) closures may be assumed.

4. Household block — household income from factor payments and transfers, and consumption demand derived from utility maximisation (e.g., Stone–Geary or CES utility functions) subject to the budget constraint.

5. Government block — tax and transfer instruments (direct and indirect taxes, tariffs, subsidies) and government revenue–expenditure accounts, with a fiscal closure rule.

6. Foreign trade block — exports and imports modelled through the Armington assumption (imperfect substitutability between domestic and foreign goods) and the small-country or large-country trade closure.

7. Equilibrium conditions and closure rules — market-clearing conditions for all commodity and factor markets, plus macro closures for the government budget, savings–investment balance, and the current account.

8. Solution algorithm — the system of nonlinear equations is solved numerically (e.g., GAMS with a solver such as GAMS/MPSGE) to obtain counterfactual equilibrium, which is then compared with the base-run equilibrium.

The statement cannot hold for all real $a,b$ (it is not true for negative numbers), but it is true when $a>b>0$, which is the case to prove.

Why the general case fails. Take $a=1,\ b=-2,\ c=2$. Then $a>b$ and $c>0$, but $a^c=1<4=b^c$. Also, for non-integer $c$, $b^c$ may not even be real when $b<0$. So the result needs $a>b>0$ (or at least $a>b\ge0$).

Proof for $a>b>0,\ c>0$. Let $f(x)=x^c$ for $x>0$. Then

$f'(x)=c\,x^{c-1}>0,$

because $c>0$ and $x^{c-1}>0$ for $x>0$. Hence $f$ is strictly increasing on $(0,\infty)$. Since $a>b>0$, $f(a)>f(b)$, that is

$a^c>b^c.$

Alternative proof (without calculus). Write $a^c=e^{c\ln a}$ and $b^c=e^{c\ln b}$. Since $\ln$ is increasing, $a>b>0$ gives $\ln a>\ln b$. Multiplying by $c>0$ keeps the inequality, and $e^{x}$ is increasing, so $e^{c\ln a}>e^{c\ln b}$. $\blacksquare$

The Solow growth model (Solow–Swan model, 1956) is the foundational neoclassical model of long-run economic growth. It explains how output per worker is determined by capital accumulation, labour growth and technological progress, and shows that only technological progress can sustain long-run growth in living standards.

Assumptions

• Production function with constant returns to scale: $Y = F(K, L) = K^{\alpha}(AL)^{1-\alpha}$ where (Y) is output, (K) capital, (L) labour and (A) the level of technology (labour-augmenting).
• Saving is a constant fraction (s) of income: (S = sY); investment (I = sY).
• Capital depreciates at a constant rate (\delta).
• Labour force grows at rate (n) and technology at rate (g).
• Perfect competition, so factors are paid their marginal products.

Capital accumulation

In intensive (per effective worker) form, with (k = K/AL) and (y = Y/AL = k^{\alpha}):

$\dot{k} = s k^{\alpha} - (n + g + \delta)k$

• (s k^{\alpha}) is actual investment per effective worker.
• ((n + g + \delta)k) is break-even investment — the investment needed to keep (k) constant as capital is diluted by population growth, technological progress and depreciation.

At the steady state, (\dot{k} = 0), giving:

$k^{*} = \left(\frac{s}{n + g + \delta}\right)^{\frac{1}{1-\alpha}}, \qquad y^{*} = (k^{*})^{\alpha}$

Because of diminishing returns to capital, a higher saving rate raises (k^{}) and (y^{}) only in levels, not in the long-run growth rate. Once (k = k^{*}), output per effective worker stops growing. This is the model's central prediction of conditional convergence: poorer countries with lower (k) grow faster and catch up to their own steady states.

Labour growth

A rise in (n) steepens the break-even investment line, lowering (k^{}) and (y^{}). Higher population growth therefore reduces steady-state income per worker (a "dilution" effect), though it raises the growth rate of total output (Y) by (n + g). This helps explain why many developing countries with high (n) remain poor.

Technological progress

Technology (g) is the only source of sustained growth in output per worker. In steady state:

• (k) and (y) are constant (per effective worker),
• capital per worker (K/L) and output per worker (Y/L) grow at rate (g),
• total output (Y) grows at rate (n + g).

Thus long-run living standards rise only through technological progress; capital accumulation alone cannot do so because of diminishing returns.

Policy implications

• Raising (s) or lowering (\delta) increases steady-state income but not its long-run growth rate.
• Lowering (n) (e.g. through family planning, education) raises income per worker.
• The only route to permanently rising per-capita income is continuous technological progress through R&D, education and innovation.

Limitation: the model treats (g) as exogenous and does not explain what drives technology — a gap later addressed by endogenous growth theory (Romer, Lucas).

a) Optimal control theory with the Hamiltonian condition

Optimal control theory finds the time path of a control variable that optimises an objective functional subject to a differential equation governing the state variable. It is used in growth theory, resource economics and optimal saving.

Problem: maximise $J=\displaystyle\int_0^T f(t,y,u)\,dt$ subject to $\dot y=g(t,y,u)$, $y(0)=y_0$, where $y$ is the state variable and $u$ the control.

Hamiltonian (present-value form):

$H(t,y,u,\lambda)=f(t,y,u)+\lambda(t)\,g(t,y,u)$

with $\lambda(t)$ the costate variable (shadow price of the state).

Necessary conditions (Pontryagin's maximum principle):

1. Control: $\max_u H$ at every $t$; for an interior solution $\partial H/\partial u=0$.
2. Costate: $\dot\lambda=-\partial H/\partial y$.
3. State: $\dot y=\partial H/\partial\lambda=g$.
4. Transversality: $\lambda(T)=0$ if $y(T)$ is free (or $H(T)=0$ if $T$ is free).

If the objective has a discount factor, $f=e^{-\rho t}F(\cdot)$, the current-value Hamiltonian $H_c=F+\mu g$ (with $\mu=\lambda e^{\rho t}$) is often used. Its conditions become $\dot\mu=\rho\mu-\partial H_c/\partial y$ and the transversality condition $\lim_{t\to\infty}e^{-\rho t}\mu(t)y(t)=0$ (or $\mu(T)e^{-\rho T}=0$ for free $y(T)$).

Interpretation: $H$ balances the current gain $f$ against the future value of the change in the state, $\lambda g$. Applications include the Ramsey optimal-saving model and Hotelling's rule for exhaustible resources.

b) Market inventory model

Producers hold inventories as a buffer between production and sales, so price adjusts according to the stock rather than clearing the market instantly.

Let $Q_d=a-bP$ (demand), $Q_s=-c+dP$ (supply), and let inventory change be $\dot I=Q_s-Q_d$. If price adjusts in the direction opposite to the inventory stock, $\dot P=-k\,(I-\bar I)$ with $k>0$ and $\bar I$ the desired stock, then excess supply accumulates as stock and price falls until the stock returns to the desired level.

In equilibrium $\dot I=0$ (so $Q_s=Q_d$) and $I=\bar I$. The inventory buffer smooths price fluctuations: excess supply is absorbed into stock instead of causing an immediate price crash. The model is relevant for storable goods such as grains, and underlies buffer-stock policies.

c) Homogeneous production function

A production function $Q=f(L,K)$ is homogeneous of degree $k$ if $f(tL,tK)=t^k f(L,K)$ for all $t>0$.

• $k=1$: constant returns to scale (linear homogeneity).
• $k>1$: increasing returns to scale.
• $k<1$: decreasing returns to scale.

Euler's theorem: $L\,f_L+K\,f_K=k\,f$. For $k=1$ the total product is exactly exhausted when each factor is paid its marginal product (product exhaustion), the basis of the marginal productivity theory of distribution.

Cobb–Douglas $Q=AL^{\alpha}K^{\beta}$ is homogeneous of degree $\alpha+\beta$. For linearly homogeneous functions, average and marginal products depend only on the capital–labour ratio $K/L$, which is why they are central in growth theory.

Formulating a Research Problem in Economics

A research problem is a clear, specific statement of the gap between what is known and what needs to be known in a chosen area. Formulating it is the first and most decisive step of research, because a badly framed problem wastes the entire study. The steps I would follow are set out below, with examples drawn from an area of my interest — labour economics in Nepal, specifically youth out-migration and its effect on agricultural productivity.

Step 1: Identify a broad area of interest

I begin with a field I know and care about. Interest sustains the long effort a study demands. My area: the economics of migration and rural labour markets in Nepal.

Step 2: Preliminary review of literature and data

Before narrowing, I read existing studies (journal articles, dissertations, government reports such as the Nepal Labour Force Survey, NLSS, Census, and Nepal Rastra Bank surveys) to learn what has already been established. This tells me where knowledge is thin. I find that much work exists on remittances and consumption, but little on how the absence of young workers affects farm output at the household level.

Step 3: Narrow the area into a specific topic

A broad area cannot be researched; it must be narrowed by place, time, population, and variable. From "migration and agriculture" I narrow to: the effect of youth out-migration on agricultural productivity in Nawalparasi district, 2015–2023.

Step 4: State the problem precisely (problem statement)

I convert the topic into a question that research can answer. For example:

"Has the out-migration of male youth aged 20–40 reduced household agricultural output per hectare in Nawalparasi, and through which channel — labour shortage, shift to less labour-intensive crops, or land abandonment?"

This states the gap: existing studies measure remittance income but not the labour-loss effect on productivity.

Step 5: Define objectives

Objectives must be specific, measurable, and few. For my study:

1. To measure the change in agricultural productivity of migrant-sending households.
2. To identify the channels through which labour loss affects output.
3. To suggest policy measures for labour-substituting technology.

Step 6: Formulate research questions and hypotheses

Each objective becomes a question, and each question a testable hypothesis:

• H₀: Out-migration has no significant effect on output per hectare.
• H₁: Out-migration significantly reduces output per hectare.

Step 7: Define scope and delimitation

I state clearly what is included and excluded — one district, one crop season, households with at least one migrant, the period 2015–2023 — so the findings are not over-generalised.

Step 8: Assess feasibility (data, time, cost, ethics)

I check whether data are obtainable. If secondary data are insufficient, I plan a primary survey of, say, 200 households using a structured questionnaire. I confirm the budget, time, and ethical clearance for interviewing migrants' families.

Step 9: Define key concepts and variables

Operational definitions prevent confusion:

• Out-migration = a household member aged 20–40 working abroad for 6+ months.
• Agricultural productivity = output (kg) per hectare per season.
• Independent variables: number of migrants, remittance received, land size, irrigation, mechanisation.

Step 10: State the significance of the study

I explain who benefits — policymakers designing labour-substitution programmes, local governments, and researchers — and how the study fills the identified gap.

Step 11: Finalise the title

A good title is short, specific, and reflects variables, place, and time:

"Youth Out-Migration and Agricultural Productivity: A Household-Level Study of Nawalparasi District, Nepal (2015–2023)."

Conclusion

The sequence — interest → literature → narrowing → problem statement → objectives → hypotheses → scope → feasibility → definitions → significance → title — converts a vague interest into a researchable problem. Each step disciplines the next, and together they ensure the study is original, feasible, and useful. For my chosen area, this process yields a problem that is economically meaningful: whether Nepal's remittance economy is quietly eroding its own food-producing base.

Types of Sampling Techniques

Sampling is the process of selecting a representative subset (sample) from a population for study. The techniques are broadly divided into two categories:

1. Probability (Random) Sampling

In this method, every unit of the population has a known, non-zero chance of being selected. It is based on the principle of randomisation and allows the calculation of sampling error.

(a) Simple Random Sampling Every unit has an equal probability of selection. The sample is drawn by lottery method or random number tables. It is simple and unbiased but requires a complete sampling frame and is costly for large, scattered populations.

(b) Systematic Sampling Units are selected at a fixed interval (k = N/n) after a random start. It is convenient and quick, but may introduce bias if the population has a periodic pattern.

(c) Stratified Random Sampling The population is divided into homogeneous strata (e.g., income groups, regions), and random samples are drawn from each stratum. It increases precision and ensures representation of all groups. It is widely used in socio-economic surveys.

(d) Cluster Sampling The population is divided into clusters (e.g., wards, villages), and entire clusters are randomly selected and studied. It is cost-effective for geographically dispersed populations but has higher sampling error.

(e) Multistage Sampling Sampling is done in successive stages (e.g., district → ward → household). It is practical for large national surveys.

2. Non-Probability (Non-Random) Sampling

Here, selection depends on the researcher's judgement or convenience, and the probability of selection is unknown.

(a) Convenience Sampling — Units are chosen because they are easily available. Quick and cheap but highly biased.

(b) Purposive (Judgement) Sampling — The researcher deliberately selects units believed to be representative or relevant.

(c) Quota Sampling — The population is divided into categories and a fixed quota is filled from each, but selection within quota is non-random.

(d) Snowball Sampling — Existing respondents refer further respondents; useful for hidden or rare populations.

(e) Voluntary Sampling — Respondents self-select, as in online polls.

---

Sampling Technique Adopted in an Experimental Study Design

In an experimental study design, the most appropriate sampling technique is random probability sampling — specifically simple random sampling (or stratified random sampling where heterogeneity exists) for the selection of subjects, followed by random assignment of subjects to treatment and control groups.

Reasons

1. Minimises selection bias Random selection gives every unit an equal chance of being included, so the sample is representative of the population. Systematic differences between subjects are avoided.

2. Ensures internal validity Random assignment (randomisation) balances both known and unknown confounding variables across treatment and control groups. This is the core requirement of a true experiment, as it allows any observed difference in the outcome to be attributed to the treatment rather than to pre-existing differences.

3. Permits statistical inference Probability sampling provides a known sampling distribution, so sampling error can be estimated. This makes it valid to apply t-tests, ANOVA, or regression and to generalise results from the sample to the population with a measurable degree of confidence.

4. Satisfies the assumption of independence Random selection and assignment ensure that observations are independent of one another, which is a basic assumption of most parametric statistical tests.

5. Controls extraneous variation If the population is heterogeneous (e.g., by income, gender, or location), stratified random sampling is used so that each stratum is represented proportionally. This reduces variance and increases the precision (power) of the experiment to detect a true treatment effect.

6. Enhances external validity (generalizability) Because the sample is drawn randomly from a defined population, the causal findings of the experiment can be generalized beyond the sample to the wider population.

Conclusion

Thus, in an experimental study design, random probability sampling (simple or stratified) combined with random assignment is adopted, because it eliminates selection bias, controls confounding factors, satisfies the assumptions of statistical inference, and strengthens both the internal and external validity of the causal conclusions drawn from the experiment.

Descriptive and Inferential Data Analysis

Descriptive Data Analysis

Definition: Descriptive data analysis is the branch of statistics concerned with collecting, organizing, summarizing, and presenting data in a meaningful form so that the basic features of a dataset become clear. It describes what is in the data without drawing conclusions beyond the data itself. It uses measures such as frequency distributions, percentages, measures of central tendency (mean, median, mode), measures of dispersion (range, variance, standard deviation, coefficient of variation), and graphical tools such as bar charts, histograms, pie charts, and scatter diagrams.

Purpose: To reduce a large mass of raw data into a few understandable summary values and to reveal patterns, trends, and comparisons within the observed data.

Inferential Data Analysis

Definition: Inferential data analysis is the branch of statistics concerned with drawing conclusions, predictions, or generalizations about a population on the basis of information contained in a sample drawn from that population. It uses probability theory to measure the reliability of those conclusions. Its main tools are estimation (point and interval estimation) and hypothesis testing (z-test, t-test, chi-square test, F-test, ANOVA, regression analysis).

Purpose: To generalize sample findings to the whole population, to test theories or claims, and to support decision-making under uncertainty.

Distinction between Descriptive and Inferential Data Analysis

Conclusion

The two are complementary rather than competing. Descriptive analysis is the first step that organizes and simplifies data, while inferential analysis builds on these summaries to make generalizations about the population. In economic research, descriptive statistics describe the sample or economy, and inferential statistics test hypotheses and forecast economic behaviour.

Research design is the overall plan or blueprint that a researcher prepares before actually carrying out a study. It specifies the type of study, the population and sample, the methods of data collection, the tools of analysis, and the time and cost involved. Its central purpose is to link the research question to the data and the conclusions in a logically sound way. The statement that research design "ensures that the procedures required to complete the study are adequate to obtain valid, objective, and accurate answers to the research question" can be explained as follows.

1. It makes the procedures adequate to the research question. A research question may be descriptive, explanatory, or exploratory. Each type demands a different design. A descriptive question ("What is the level of financial literacy among rural households?") needs a survey design with a representative sample; an explanatory question ("Does microcredit reduce poverty?") needs a causal or quasi-experimental design with control groups. By choosing the design that matches the question, the researcher ensures that the procedures adopted are capable of answering that particular question rather than some other question. Without a design, the researcher may collect data that are interesting but irrelevant to the objective.

2. It ensures validity. Validity means that the study measures what it claims to measure and that the conclusions follow from the evidence. Research design safeguards validity in several ways:

• Internal validity: Random sampling and random assignment, use of control groups, and control of extraneous variables reduce bias and confounding, so that an observed effect can be attributed to the treatment rather than to other factors.
• External validity: A clearly defined population and an appropriate sampling frame allow the findings to be generalized beyond the sample.
• Construct validity: Clear operational definitions of concepts (for example, defining "poverty" by a consumption expenditure line) ensure that the indicators actually represent the theoretical construct.
• Statistical conclusion validity: Choosing an adequate sample size and the correct statistical test reduces Type I and Type II errors.

3. It ensures objectivity. Objectivity means that the findings do not depend on the personal bias, convenience, or preconceptions of the researcher. A well-prepared design achieves this by pre-specifying the sampling method (probability sampling rather than convenience sampling), standardising the instruments (structured questionnaires, uniform instructions to enumerators), pre-testing the tools, and fixing the hypotheses and analysis plan in advance. When procedures are predetermined and applied uniformly to all units, the personal judgement of the investigator plays a minimal role, and another researcher repeating the study should reach similar results.

4. It ensures accuracy. Accuracy refers to the closeness of the estimates to the true population values and the reliability of the measurements. Design contributes to accuracy through:

• correct definition of population, unit of analysis, and variables;
• use of reliable and pre-tested instruments;
• adequate sample size and low sampling error;
• proper training and supervision of data collectors to reduce non-sampling error;
• appropriate data processing and use of suitable analytical techniques.

5. It provides a basis for control and economy. By deciding in advance what data are needed, from whom, how, and when, the design prevents the collection of superfluous data. This saves time, money, and effort, and keeps the study within feasible limits — an important consideration in economics research where surveys are costly.

6. It guides the entire research process. The design acts as a roadmap: it tells the researcher how to proceed at each stage, from problem formulation to sampling, measurement, analysis, and interpretation. This reduces confusion, prevents mid-course changes that could introduce bias, and makes the study replicable and comparable with other studies.

Conclusion. A well-constructed research design is therefore not a mere formality but the foundation of good research. It aligns the research question with suitable procedures, and by doing so it secures the three essential qualities of scientific enquiry — validity, objectivity, and accuracy. A defective design, however carefully the data are later collected and analysed, cannot be repaired at the analysis stage; hence the saying that the whole research stands or falls on its design.

Evaluating Scientific Literature and Writing a Literature Review

A literature review is a critical, systematic summary and synthesis of existing knowledge on a research question. It is not a list of summaries but an argument that shows what is known, what is contested, and what remains unstudied. It therefore requires both evaluation of individual studies and organisation of the evidence as a whole.

A. Evaluating Scientific Literature

Each source is judged on the following criteria:

1. Relevance – Does the study address the research question, the same population, period and context (e.g., a study of inflation in Nepal vs. one in the USA)?
2. Authority and credibility – Is the author/institution recognised? Is the source peer-reviewed, or a working paper, report or blog?
3. Methodology – Is the design appropriate (experiment, survey, panel data, RCT)? Is the sample adequate and representative? Are the variables measured correctly? Is there a control group or counterfactual?
4. Validity and reliability – Internal validity (causal inference, omitted variable bias, endogeneity), external validity (generalisability), and consistency of results across studies.
5. Objectivity and bias – Funding source, ideological slant, selective reporting, publication bias.
6. Currency – Recent enough for the topic; classic works are retained only for theory.
7. Contribution – Does it add new evidence, method or theory, or merely repeat earlier work?

B. Steps in Writing a Literature Review

1. Define the question and scope – specify topic, time frame, geography and key concepts.
2. Search the literature – use databases (JSTOR, Google Scholar, NepJOL and EconLit), keywords, and backward/forward citation tracking.
3. Screen and select – apply inclusion/exclusion criteria; keep only relevant, quality studies.
4. Read and record – prepare a matrix of author, year, objective, method, data, findings and limitations.
5. Critically appraise – apply the criteria in Part A; note agreements and contradictions.
6. Synthesise thematically – group studies by theme, theory, method or finding rather than by author; compare and contrast; explain why results differ.
7. Identify the research gap – the unresolved issue that justifies the new study.
8. Write the review – structure it as: introduction (topic, purpose, roadmap), body (themes with evidence and critique), conclusion (summary of knowledge, gap, implications).
9. Cite and reference properly – follow APA style consistently; avoid plagiarism.

Conclusion

Good evaluation asks whether a study is relevant, credible, methodologically sound and current; good reviewing goes further by synthesising the evidence into themes and exposing the gap. A review that merely describes studies is weak; one that analyses and integrates them provides the theoretical and empirical foundation for the research.

Differences between Concept and Variables

In short, a concept is an idea; a variable is that idea made measurable so that data can be collected and tested.

Conversion of Concepts into Variables

A concept is converted into a variable through operationalisation — the process of defining an abstract concept in terms of measurable indicators. The steps are:

1. Define the concept clearly — state exactly what is meant (e.g., poverty as inability to meet basic needs).
2. Identify dimensions — break the concept into its components (e.g., income, consumption, health, education).
3. Select indicators — choose observable measures for each dimension (e.g., per capita income, calorie intake, school enrolment).
4. Assign measurement scale — decide whether the variable is nominal, ordinal, interval or ratio.
5. Fix units and categories — e.g., income in Rs per month, or poor/non-poor classification.
6. Pilot test and refine — check reliability and validity of the measure before final use.

Example: The concept "poverty" → dimension: economic deprivation → indicator: monthly per capita consumption expenditure → variable: per capita consumption expenditure in Rs per month (ratio scale), with a poverty line to classify households as poor or non-poor.

Methods of Quantitative Data Collection

Quantitative data (numerical, measurable information) is collected mainly through the following methods:

1. Survey method — structured questionnaires or interviews administered to a sample (mail, telephone, online, face-to-face).
2. Observation method — systematic recording of countable behaviour or events, often with checklists or rating scales.
3. Experiment method — controlled manipulation of variables to measure cause–effect relationships (lab or field experiments).
4. Secondary data method — use of already-published numerical records: government statistics, central bank reports, census, surveys of national/international agencies.
5. Administrative/by-product records — data generated from routine operations (school registers, hospital records, tax files).

Method I Would Follow and Why

For a study of a socio-economic issue such as household income, consumption or employment, I would follow the sample survey method using a structured questionnaire.

Reasons:

• Representativeness: A properly drawn random sample allows valid generalization to the whole population, which observation or case records cannot give.
• Standardization and comparability: Fixed questions and response options produce uniform numerical data that can be coded, tabulated and statistically analysed (mean, variance, regression).
• Cost and time efficiency: It covers a large population at far lower cost and time than a census or a controlled experiment.
• Flexibility of administration: It can be conducted face-to-face, by phone or online, and can include both closed questions (for quantitative measurement) and a few open questions for context.
• High response quality when designed well: With trained enumerators and pre-testing, problems of non-response and measurement error can be minimized.

Thus, the sample survey is the most practical, reliable and widely used method for collecting quantitative data in economics, provided the sampling design and questionnaire are carefully prepared.

Participatory Rural Appraisal (PRA)

Meaning (2 marks)

Participatory Rural Appraisal is a family of approaches and methods that enable rural people to share, enhance and analyse their own knowledge of life and conditions, to plan and to act. It emerged in the late 1980s and early 1990s, growing out of Rapid Rural Appraisal (RRA) and the participatory research traditions of Robert Chambers and others. Its defining feature is that the outside professional acts as a facilitator and learner rather than as an extractor of data: local people are treated as the primary analysts of their own situation, and the visual and verbal methods used are ones they can handle themselves. It is thus both a research approach and a planning/empowerment tool, widely used in Nepal by NGOs, donor projects, community forestry and local government planning.

How PRA works (3 marks)

1. Entry and rapport building. The team first visits the community, meets key informants, explains the purpose and seeks consent. Trust and rapport are prerequisites; the team avoids a "surveyor" posture and behaves as a guest.

2. Selection of methods and tools. A mix of flexible, visual, group-based tools is chosen according to the issue. Common tools include:

   • Social mapping — villagers draw their settlement, showing households, caste/ethnic clusters, roads, water points and services.
   • Resource mapping — land use, forest, irrigation and cropping patterns.
   • Transect walk — a systematic walk across the village with local guides to observe and discuss zones, soils and problems.
   • Seasonal calendar — cropping, labour, migration, food availability and disease by month.
   • Wealth ranking / well-being ranking — households classified by local criteria of poverty and wealth.
   • Venn diagram (institutional diagram) — local institutions and their importance shown as overlapping circles.
   • Matrix ranking and scoring — comparing options (crops, tree species, technologies) by locally chosen criteria.
   • Focus group discussions, key informant interviews, timelines and trend analysis.

3. Triangulation and iteration. Findings from one tool or group are cross-checked against others (different methods, different informants, different sites) to reduce bias. The process is iterative — hypotheses are revised as understanding deepens.

4. Analysis by the community. The outputs (maps, charts, rankings) remain with the community and are discussed in group meetings and plenary sessions, where villagers themselves identify problems, priorities and possible solutions. The outsider's role is to ask probing questions, not to supply answers.

5. Planning, action and follow-up. The analysis feeds into a Participatory Rural Appraisal plan / community action plan, which is implemented with local institutions, monitored and reviewed. The process is meant to be continuous, with the community retaining ownership of both the information and the decisions.

Key principles: reversal of learning (learning from, not about, local people), offsetting biases (spatial, wealth, gender, seasonal, professional), optimal ignorance (not collecting more than is needed), triangulation, seeking diversity rather than averages, and facilitating local analysis.

Measurement Scales

Measurement scales are the ways in which variables (numbers or symbols) are assigned to objects or events according to a set of rules. They determine what statistical operations can legitimately be performed on the data. There are four levels of measurement, arranged from the weakest to the strongest: nominal, ordinal, interval, and ratio.

1. Nominal Scale This is the simplest and weakest level. Numbers or labels are used only to name, identify, or classify objects into mutually exclusive and collectively exhaustive categories. There is no order or ranking among the categories. The numbers have no quantitative meaning — they are merely tags. Examples: gender (male = 1, female = 2), religion, marital status, blood group, district of birth. Only the mode and frequency counts (chi-square, contingency tests) can be used; arithmetic operations like addition or averaging are meaningless.

2. Ordinal Scale Here the categories are ranked or ordered according to some criterion, but the intervals between ranks are not equal or known. We can say one item is greater or better than another, but not by how much. Examples: socio-economic status (low, middle, high), Likert-scale responses (strongly disagree to strongly agree), education level, ranking of preferences. The median and mode, along with non-parametric tests such as the Mann–Whitney U or Spearman's rank correlation, are appropriate.

3. Interval Scale This scale has ordered categories with equal intervals between consecutive points, so differences between values are meaningful. However, it lacks a true or absolute zero, so ratios are not meaningful. Examples: temperature in Celsius or Fahrenheit, IQ scores, calendar years. We can say 40°C is 10°C warmer than 30°C, but not that 40°C is twice as hot as 20°C. The mean, standard deviation, correlation, and t-tests can be applied.

4. Ratio Scale This is the highest and most complete level. It possesses all the properties of the interval scale plus a true, meaningful absolute zero, which indicates the complete absence of the attribute. Hence all arithmetic operations, including multiplication and division (ratios), are valid. Examples: income, weight, height, age, number of children, GDP. Statements such as "A earns twice as much as B" are meaningful. Almost all statistical techniques are applicable.

Summary: The nominal scale only classifies, the ordinal scale classifies and ranks, the interval scale adds equal distances, and the ratio scale further adds a true zero. The choice of scale determines which statistical tools — from simple frequency counts to advanced parametric tests — can be validly used in economic and social research.

Research proposal — A research proposal is a formal, written document that a researcher prepares before conducting a study. It outlines the research problem, objectives, research questions, hypotheses, methodology, scope, time frame and budget of the proposed study. It serves as a plan or blueprint that guides the researcher and seeks approval or funding from a supervisor, department or funding agency. In economics, a proposal typically states the topic (e.g., "Determinants of remittance use in Nepal"), the rationale, the data and the econometric method to be used.

Three key functions of a research proposal

1. Planning and direction (blueprint function): It forces the researcher to think through the problem, objectives, hypotheses and methodology in advance. It specifies what data will be collected, from where, and how it will be analysed, so the study proceeds in a systematic way and avoids confusion and wasted effort.

2. Communication and justification (persuasive function): It communicates the research idea to supervisors, departments, funding agencies or ethical review committees. It justifies why the study is worth doing — its significance, originality and contribution to economic knowledge or policy — and thereby helps secure approval, funding and cooperation.

3. Evaluation and control (monitoring function): It provides a basis for evaluating the feasibility and quality of the study before it begins, and later for monitoring progress. Supervisors and funding bodies can judge whether the objectives, methods, budget and time schedule are realistic, and can check whether the completed work matches what was proposed.

Preliminaries, Main Body and Appendices of a Research Report

A research report is organised into three broad divisions, each serving a distinct purpose.

1. Preliminaries (prefatory parts) These appear before the actual text and introduce the report to the reader. They include the title page, letter of transmittal, table of contents, list of tables and figures, and acknowledgements. They are not numbered as chapters and are written briefly. Example: In a study on "Determinants of Household Saving in Kathmandu Valley," the title page, table of contents and list of tables form the preliminaries.

2. Main Body (text) This is the core of the report, containing the actual research work in a logical sequence — introduction, statement of the problem, objectives, review of literature, methodology, data presentation and analysis, findings, conclusions and recommendations. It carries the main argument and evidence, and is divided into numbered chapters or sections. Example: The same study's chapters on research methodology, regression analysis of saving behaviour, and conclusions constitute the main body.

3. Appendices (supplementary parts) These come after the main body and contain supporting material that is too detailed or bulky to include in the text but is useful for verification. They include the questionnaire, raw data tables, computer outputs, mathematical derivations and the bibliography or references. Example: The sample questionnaire, SPSS regression output and the list of references attached at the end of the saving study form the appendices.

Key difference: Preliminaries help the reader locate and understand the report, the main body presents and analyses the research findings, and the appendices support and authenticate the findings with detailed evidence.

Data cleaning: key steps to make collected data ready for analysis

Raw data collected from the field is almost never ready for analysis. It usually contains errors, missing values, inconsistencies and outliers. Data cleaning (also called data preparation or pre-processing) is the process of detecting, correcting or removing these problems so that the data becomes accurate, complete, consistent and usable. The key steps are as follows:

1. Data editing and validation The first step is to check the raw data for completeness and accuracy. Editing may be field editing (done on the spot while collecting data) or central editing (done at the office after collection). The researcher checks whether all questions have been answered, whether entries are legible, and whether the recorded values fall within a reasonable range. Invalid entries — for example, an age of 250 years or a sex code of "5" — are identified and corrected or removed.

2. Handling missing values Missing data must be dealt with because most statistical techniques cannot handle blanks. The researcher first examines the pattern of missingness: whether it is random or systematic. Common remedies are:

• Deletion: dropping the incomplete case (list-wise) or the variable (column-wise) if the loss is small.
• Imputation: replacing the missing value with the mean, median or mode; with the value from a similar respondent (hot-deck); or with a model-based estimate (regression imputation). The choice depends on the amount and nature of missingness; careless imputation can bias results.

3. Removing duplicates and correcting inconsistencies Duplicate records (common in survey and secondary data) are detected and removed. Inconsistencies across related variables are also corrected — for example, a respondent recorded as "unmarried" but having a spouse, or age and date of birth that do not match. This ensures internal consistency of the data set.

4. Identifying and treating outliers Outliers are extreme values that may be genuine or due to data-entry mistakes. They are detected using box plots, z-scores or the interquartile range rule. Genuine outliers may be retained or handled by winsorizing/transformation; erroneous ones are corrected or deleted. Outliers matter because they distort means, variances and regression estimates.

5. Data coding and transformation Responses are converted into numerical codes suitable for software (for example, male = 1, female = 0). New variables may be created through transformation — recoding into categories, computing ratios, indices or scales, and standardising units (for example, converting all income figures to Rs per month). This step makes the data machine-readable and comparable.

6. Checking data entry and structure If data were entered manually, a verification or double-entry check is done to catch typing errors. The researcher also checks the structure of the data set — variable names, data types (numeric, string, date), and the correct number of observations — so that it can be imported into software such as SPSS, STATA or R without error.

7. Final screening and documentation Finally, the cleaned data set is screened once more using descriptive statistics (frequencies, means, ranges) to confirm that all values are plausible. Every cleaning decision — deletions, imputations, recodes — should be documented in a codebook or cleaning log so that the process is transparent and reproducible.

Conclusion Data cleaning is not a single act but a systematic sequence of editing, handling missing values, removing duplicates, treating outliers, coding, transforming and verifying the data. Only after these steps is the data set accurate, consistent and complete enough for meaningful statistical analysis; skipping them risks biased and unreliable conclusions.

a) Footnotes and Endnotes

Footnotes and endnotes are supplementary notes used in research writing to provide additional information, explanations, citations, or comments without interrupting the main text.

Footnotes appear at the bottom of the same page on which the reference is made. They are separated from the main text by a short horizontal line. The reader can glance down the page and immediately see the note, so footnotes are convenient when the note is closely tied to the discussion on that page.

Endnotes appear together at the end of the chapter, article, or the whole document, usually under the heading "Notes." They keep the pages of the main text clean and uncluttered, but the reader has to turn to the end of the document to consult them.

Key points of difference:

Both are indicated in the text by a superscript number (or symbol such as *, †) placed after the relevant word or sentence. In economics research, footnotes are commonly used to explain data sources, definitions of variables, and assumptions of a model, while endnotes are used when references are numerous. Their main advantage is that they keep the argument of the main text continuous and readable while still giving the reader the supporting detail.

---

b) Research in Economics

Research in Economics is a systematic and scientific process of inquiry aimed at discovering, explaining, and predicting economic behaviour and economic phenomena. It involves the collection, organisation, analysis, and interpretation of economic data to answer a well-defined question or to test a hypothesis.

Objectives:

1. To describe economic phenomena (what is happening).
2. To explain causal relationships among economic variables (why it happens).
3. To predict future economic trends and outcomes.
4. To provide a basis for policy formulation and decision-making.

Types:

• Pure (basic) research — develops economic theory, e.g. testing the permanent income hypothesis.
• Applied research — solves practical problems, e.g. studying the impact of a subsidy on rice production.
• Descriptive vs. analytical, and quantitative vs. qualitative research are other common classifications.

Steps in economic research:

1. Formulating the research problem.
2. Reviewing the literature.
3. Formulating hypotheses.
4. Designing the research (sampling, data sources).
5. Collecting data (primary or secondary).
6. Analysing data using statistical and econometric tools.
7. Interpreting results and drawing conclusions.
8. Writing the research report.

Importance: It helps in testing economic theories, measuring relationships such as elasticity, demand, inflation and growth, evaluating government policies, and guiding planning and development in countries like Nepal. Its main limitations are the non-experimental nature of economic data, measurement errors, and the difficulty of controlling all variables.

---

c) Univariate Data Analysis

Univariate data analysis is the statistical analysis of a single variable at a time. The word "uni" means one and "variate" means variable, so the analysis deals with one characteristic of a set of observations — for example, the monthly income of 100 households, or the marks of 50 students in economics.

Purpose: To describe the central tendency, dispersion, and shape of the distribution of that one variable, and to detect outliers.

Main tools:

1. Measures of central tendency — mean, median, mode. They show the typical or average value of the variable.
2. Measures of dispersion — range, quartile deviation, mean deviation, variance, standard deviation, coefficient of variation. They show how spread out the values are.
3. Measures of shape — skewness and kurtosis, which describe asymmetry and peakedness of the distribution.
4. Graphical presentation — frequency tables, histograms, bar diagrams, pie charts, box plots, stem-and-leaf plots.
5. Distribution fitting and testing — checking whether the data follow a normal or other theoretical distribution, using tests such as chi-square goodness of fit.

Uses in economics: describing income distribution, poverty and inequality (Lorenz curve, Gini coefficient), analysing price and consumption patterns, summarising survey data, and preparing the ground for bivariate and multivariate analysis.

Limitations: It ignores relationships between variables. A univariate analysis of income alone cannot show how income depends on education or family size; for that, bivariate or multivariate techniques are needed.

Measurement in Research

Measurement is the process of assigning numerals or other symbols to the properties or characteristics of objects, persons, or events according to a set of rules. In research, measurement provides the quantitative or qualitative description of variables so that they can be analysed, compared, and interpreted systematically. For example, assigning the number "1" to male and "2" to female respondents is a measurement of the variable "gender". A good measurement must be valid (it measures what it is intended to measure) and reliable (it gives consistent results on repeated application).

Measurement in social science research is more difficult than in physical sciences because many variables (attitudes, satisfaction, motivation) are abstract and cannot be measured directly; they are measured through indicators or proxy variables.

---

Types of Measurement Scales

The four levels of measurement, arranged from the lowest to the highest order of precision, are as follows:

1. Nominal Scale

This is the simplest and most basic scale, in which numbers or labels are used only to name, identify or classify objects into mutually exclusive and collectively exhaustive categories. The numbers have no quantitative meaning — no order, no distance, no origin. Only the mode can be computed, and the appropriate statistics are frequency, percentage, and chi-square.

Example: Classification of respondents by religion — 1 = Hindu, 2 = Muslim, 3 = Buddhist, 4 = Christian; or by gender — 1 = Male, 2 = Female. Here "2" is not greater than "1"; the numerals are mere labels.

2. Ordinal Scale

This scale ranks or orders the objects according to the degree to which they possess a particular attribute. It tells us whether one item is greater or smaller than another, but not by how much. The intervals between ranks are unequal and unknown. Median, mode, percentile, and rank-order correlation can be used.

Example: Socio-economic status of households ranked as 1 = Low, 2 = Middle, 3 = High; or a Likert-scale response — Strongly Disagree, Disagree, Neutral, Agree, Strongly Agree. We know the order but not the exact distance between "Low" and "Middle".

3. Interval Scale

This scale has the properties of order and equal distance between consecutive points, so that the difference between any two values is meaningful. However, it has an arbitrary (non-absolute) zero; therefore ratios cannot be interpreted meaningfully. Arithmetic mean, standard deviation, and correlation can be computed.

Example: Measurement of temperature in Celsius or Fahrenheit — the difference between 20°C and 30°C equals that between 30°C and 40°C, but 40°C is not "twice as hot" as 20°C because 0°C does not mean the absence of temperature. Similarly, IQ scores and calendar years are interval data.

4. Ratio Scale

This is the highest level of measurement. It possesses all the properties of the interval scale plus an absolute or true zero, which represents the complete absence of the attribute. Hence all arithmetic operations, including multiplication and division, are meaningful. Geometric mean and coefficient of variation can also be used.

Example: Measurement of income, age, weight, height, or number of children. A person with an income of Rs 60,000 earns exactly twice the income of a person earning Rs 30,000, and zero income means no income at all.

---

Summary Table

Conclusion: The choice of measurement scale determines the statistical techniques applicable to the data. A researcher must therefore select the appropriate scale in accordance with the nature of the variable and the objectives of the study.

Major Components of a Research Proposal

A research proposal is a written plan that outlines what the researcher intends to study, why it is worth studying, and how it will be carried out. In economics, it serves both as a guide for the researcher and as a document for obtaining approval, funding, or academic supervision. The major components are described below.

1. Title of the Study

The title should be clear, specific, and self-explanatory. It should indicate the subject matter, the variables involved, and the scope of the study. A good title is concise and reflects the core research problem.

2. Introduction and Background

This section presents the context of the study. It explains the broader economic setting, the historical or theoretical background of the problem, and the relevance of the topic. It gradually narrows down from the general context to the specific area of inquiry.

3. Statement of the Problem

Here the researcher states precisely what problem is being investigated. It identifies the gap between the existing situation and the desired situation, points out deficiencies in existing knowledge, and formulates the issue in clear, researchable terms. A well-stated problem clarifies the direction of the entire study.

4. Objectives of the Study

The objectives state what the researcher intends to achieve. They should be specific, measurable, and consistent with the problem. A proposal usually has one general objective and a few specific objectives that break it into operational parts.

5. Research Questions or Hypotheses

Research questions are the specific queries the study seeks to answer. Where the study is explanatory or theory-testing, these are converted into testable hypotheses—statements about the expected relationship between variables that can be accepted or rejected on the basis of evidence.

6. Rationale and Significance of the Study

This section explains why the study matters. It states the academic contribution (filling a knowledge gap), the policy relevance (guiding decision-makers), and the practical usefulness for institutions, planners, or society at large.

7. Literature Review

The proposal reviews relevant theoretical and empirical literature. It summarises earlier studies, identifies their findings, methods, and limitations, and shows how the present study differs from or extends them. This establishes the theoretical foundation and avoids duplication.

8. Conceptual and Theoretical Framework

The researcher presents the concepts, variables, and their expected relationships, often in the form of a diagram or model. This framework links the research problem to established economic theory and guides the selection of variables and methods.

9. Research Methodology

This is the core of the proposal and typically includes:

• Research design — descriptive, analytical, explanatory, or exploratory.
• Nature and sources of data — primary (survey, interview) or secondary (CBS, NRB, World Bank reports).
• Sampling design — population, sample size, and sampling technique.
• Methods of data collection — questionnaire, interview schedule, observation, or documentary sources.
• Methods of data analysis — descriptive statistics, regression, correlation, or other econometric tools.
• Model specification — the functional form of the model and definition of variables.

10. Scope and Limitations of the Study

The scope defines the geographical area, time period, and subject coverage. The limitations acknowledge constraints such as time, budget, data availability, and the assumptions under which the findings will hold.

11. Chapter Plan or Organisation of the Study

This outlines how the final report will be structured—introduction, literature review, methodology, results and discussion, and conclusion with recommendations.

12. Work Schedule (Time Schedule)

A time frame, often presented in a Gantt chart, showing the sequence and duration of each activity from proposal writing to final report submission.

13. Budget Estimate

An estimate of the financial requirements—stationery, travel, data collection, data processing, and miscellaneous costs—usually presented in a table.

14. References

A list of all sources cited in the proposal, arranged alphabetically and formatted in a standard style such as APA.

Conclusion

These components together make a proposal systematic, feasible, and academically sound. A well-prepared proposal reduces uncertainty during research, ensures efficient use of resources, and provides a clear basis for evaluating the completed study.

Descriptive Method of Data Analysis

Meaning. The descriptive method of data analysis is the set of statistical techniques used to organise, summarise and present data in a compact and meaningful form so that the basic features of the data can be understood without drawing inferences beyond the collected data. It answers the question "what is in the data?" rather than "what can we conclude about the population?" It is the first and essential step of any economic analysis, because data must be cleaned and summarised before inferential or causal analysis is attempted.

Objectives.

1. To reduce a large mass of raw observations into a few meaningful figures.
2. To reveal the central tendency, dispersion, shape and interrelationships present in the data.
3. To make comparison between groups, regions or time periods possible.
4. To present the findings in tables, graphs and summary numbers for policy use.

Main tools of the descriptive method.

(a) Measures of central tendency. These locate the typical value of the series:

• Arithmetic mean, $\bar{X} = \dfrac{\sum X}{N}$ (or $\dfrac{\sum fX}{\sum f}$ for grouped data)
• Median — the positional middle value, useful for skewed distributions
• Mode — the most frequently occurring value, useful for qualitative and discrete data
• Geometric and harmonic means — used for growth rates and index numbers

(b) Measures of dispersion. These show how far the observations spread around the average:

• Range and quartile deviation
• Mean deviation
• Variance, $\sigma^2 = \dfrac{\sum (X-\bar{X})^2}{N}$
• Standard deviation, $\sigma = \sqrt{\dfrac{\sum (X-\bar{X})^2}{N}}$
• Coefficient of variation, $CV = \dfrac{\sigma}{\bar{X}} \times 100$, used to compare variability across groups with different means

(c) Measures of shape. Skewness (asymmetry of the distribution) and kurtosis (peakedness) describe the form of the frequency curve. Karl Pearson's coefficient of skewness is $Sk = \dfrac{\bar{X} - M_o}{\sigma}$.

(d) Measures of relationship. Simple correlation coefficient, $r = \dfrac{N\sum XY - (\sum X)(\sum Y)}{\sqrt{[N\sum X^2 - (\sum X)^2][N\sum Y^2 - (\sum Y)^2]}}$, and simple linear regression measure the degree and direction of association between two variables.

(d) Tabular and graphical presentation. Frequency distributions, cross-tabulations, bar diagrams, pie charts, histograms, frequency polygons, ogives and time-series line graphs present the data visually. In economics, Lorenz curves and scatter diagrams are also used.

(e) Index numbers and growth rates. Wholesale and consumer price indices, GDP deflators and compound growth rates summarise changes over time in a single figure.

Steps in the descriptive method.

1. Data collection — from primary sources (survey, field observation) or secondary sources (CBS, NRB, World Bank).
2. Editing and cleaning — checking for omissions, inconsistencies and outliers.
3. Classification and tabulation — arranging data by attribute, time, or class interval.
4. Computation of summary measures — averages, dispersion, correlation, index numbers.
5. Presentation — tables, diagrams and graphs.
6. Interpretation — drawing simple conclusions about the pattern, level and variability of the data.

Uses in economics.

• Measuring national income, inflation, unemployment and poverty levels.
• Comparing per capita income, HDI or consumption across provinces and countries.
• Studying the trend and seasonal pattern of prices, remittances, exports and agricultural output.
• Providing the factual base for planning, budgeting and policy formulation.
• Supplying the input data for inferential statistics (estimation and hypothesis testing).

Limitations.

1. It describes only the collected sample or series; it cannot generalise to the population.
2. It cannot establish cause and effect — correlation is not causation.
3. Results depend heavily on the quality, completeness and representativeness of the data.
4. Summary measures may hide important details and are sensitive to extreme values.
5. It is not suitable for testing hypotheses or for forecasting.

Conclusion. The descriptive method is the foundation of economic data analysis. By converting raw data into averages, measures of dispersion, tables and graphs, it gives a clear picture of the level, variability and trend of economic variables. It must, however, be supplemented by inferential and econometric methods when the objective is generalisation, hypothesis testing or causal explanation.

Significance of Research in Economics

Research occupies a central place in economics, both as a scientific discipline and as a guide to policy. Its significance can be accounted under the following points:

1. Development of economic theory. Economic theories are not speculative constructions; they are built and tested through research. The classical, Keynesian, and monetarist revolutions in economic thought each emerged from systematic observation and analysis of real economic phenomena. Research thus supplies the empirical foundation on which theory stands or falls.

2. Formulation of economic policy. Sound policy requires evidence. Whether the issue is inflation control, taxation reform, employment generation, or poverty reduction, planners and policymakers depend on research findings to choose instruments, estimate their effects, and avoid unintended consequences. Research converts political intentions into workable economic measures.

3. Solution of economic problems. Problems such as unemployment, inequality, balance-of-payments deficits, food insecurity, and energy crisis are complex and multidimensional. Research identifies their causes, measures their magnitude, and evaluates alternative remedies, thereby enabling rational rather than ad hoc decision-making.

4. Measurement of economic variables and performance. National income accounting, price indices, poverty lines, human development indices, and growth rates all rest on research-based methodology. Reliable measurement is a precondition for comparison over time and across countries, and for judging whether an economy is progressing or regressing.

5. Formulation and evaluation of development plans. In a developing economy like Nepal, periodic plans, budget allocations, and donor-supported programmes require feasibility studies, baseline surveys, and impact evaluations. Research informs project selection, resource allocation, and mid-course correction.

6. Business and managerial decision-making. Firms use market research, demand forecasting, cost analysis, and consumer surveys to decide on pricing, production, investment, and marketing. Research reduces business uncertainty and improves competitiveness.

7. Testing and refinement of economic hypotheses. Through quantitative techniques such as regression, time-series analysis, and econometric modelling, research verifies whether postulated relationships hold in practice. This makes economics a progressive, self-correcting science rather than a body of fixed dogma.

8. Understanding of economic behaviour. Research in microeconomics, behavioural economics, and institutional economics explains how consumers, producers, households, and institutions actually behave, which is essential for designing effective incentives and regulations.

9. International comparison and learning. Comparative research allows countries to learn from one another's successes and failures, and provides the evidence base for international trade negotiations, foreign aid, and global economic cooperation.

10. Academic and professional development. Research generates new knowledge, trains economists in analytical skills, and supplies the literature on which teaching, further study, and professional practice depend.

Conclusion. Research is therefore not an optional academic luxury but the backbone of economic science and economic policy. It provides the facts without which theory is empty and policy is blind, and it is indispensable for a developing economy seeking sustained, inclusive, and evidence-based growth.

Major Types of Research Design in Social Research

A research design is the overall plan or blueprint that guides the collection, measurement, and analysis of data. In social research, designs are broadly grouped into three major types according to the purpose of the study.

1. Exploratory (Formulative) Research Design Used when the problem is new, vague, or little understood. Its aim is to gain familiarity with the phenomenon and generate hypotheses rather than test them. Common methods include literature review, case studies, key informant interviews, focus group discussions, and pilot surveys. It is flexible, unstructured, and relies mainly on qualitative data. Example: an exploratory study on the livelihood strategies of climate migrants in Karnali.

2. Descriptive Research Design Used to describe the characteristics of a population, situation, or event as it exists, answering the questions what, who, where, when, and how much. It involves careful definition of the population, sample, and variables, and uses surveys, censuses, correlational studies, and comparative studies. It does not manipulate variables, so it cannot establish causation. Example: a survey describing the level of financial literacy among rural households in Nepal.

3. Explanatory (Causal or Experimental) Research Design Used to explain why relationships exist and to test causal hypotheses. It is of two kinds:

• Experimental design — the researcher manipulates the independent variable and controls extraneous factors, using control and treatment groups with random assignment (e.g., a randomised field experiment on the effect of a cash transfer on school enrolment).
• Quasi-experimental and non-experimental causal designs — used when randomisation is not possible; causal inference is drawn through statistical control, panel data, instrumental variables, or difference-in-differences (e.g., impact evaluation of a microcredit programme).

Supplementary classification

• Exploratory, descriptive, and explanatory — based on purpose.
• Cross-sectional and longitudinal — based on the time dimension; longitudinal designs include trend, cohort, and panel studies.
• Quantitative and qualitative — based on the nature of data; mixed-method designs combine both.

In practice, these types form a continuum: exploratory work generates hypotheses, descriptive work measures them, and explanatory work tests them. The choice depends on the research question, objectives, available time and resources, and the nature of the problem being studied.

Procedure for the $\chi^2$ (Chi-square) Test

The $\chi^2$ test is a non-parametric test used to examine whether observed frequencies differ significantly from expected frequencies. Its procedure involves the following steps:

1. Set up the hypotheses

• Null hypothesis ($H_0$): There is no significant difference between observed and expected frequencies (the variables are independent / the data fits the claimed distribution).
• Alternative hypothesis ($H_1$): There is a significant difference between observed and expected frequencies.

2. Choose the level of significance ($\alpha$) Usually $\alpha = 0.05$ or $0.01$ is selected, along with the appropriate degrees of freedom.

3. Calculate the expected frequencies ($E$) Expected frequency for each cell is obtained as:

$E = \frac{(\text{Row total}) \times (\text{Column total})}{\text{Grand total}}$

4. Compute the value of $\chi^2$ Using the formula:

$\chi^2 = \sum \frac{(O - E)^2}{E}$

where $O$ = observed frequency and $E$ = expected frequency. For each cell, the deviation $(O-E)$ is squared, divided by $E$, and all such values are summed.

5. Determine the degrees of freedom (df)

• For a goodness-of-fit test: $df = n - 1$ (or $n - k - 1$ when parameters are estimated).
• For a contingency table: $df = (r - 1)(c - 1)$, where $r$ = number of rows and $c$ = number of columns.

6. Find the critical (table) value of $\chi^2$ From the $\chi^2$ table, at the chosen $\alpha$ and computed $df$, the critical value $\chi^2_{\alpha, df}$ is located.

7. Compare and decide

• If calculated $\chi^2 >$ table $\chi^2$, reject $H_0$ (significant difference exists).
• If calculated $\chi^2 <$ table $\chi^2$, accept $H_0$ (no significant difference).

8. Conclusion The result is interpreted in the context of the problem, stating whether the observed data are consistent with the expected pattern or whether the variables are associated.

Note: The test assumes that expected frequencies are not too small (generally $E \geq 5$ for each cell); otherwise, adjacent categories should be pooled.

Factors to Consider While Using Focus Group Discussion (FGD) for Qualitative Data Collection

1. Purpose and research question — FGD is suitable when the aim is to explore shared experiences, group norms, perceptions, and attitudes rather than individual biographies. It should not be used where sensitive personal or confidential matters are involved.

2. Composition of the group (homogeneity vs. heterogeneity) — Participants should be similar enough (age, sex, caste/ethnicity, occupation, education) to feel comfortable, but different enough to generate diverse views. Mixing power levels (e.g., a boss with subordinates) suppresses free expression.

3. Group size — Ideally 6–10 participants. Too few limits discussion; too many prevents everyone from speaking and makes moderation difficult.

4. Number of groups (saturation) — Usually 3–6 FGDs per category of participants, continued until new themes stop emerging (theoretical saturation).

5. Selection and recruitment of participants — Purposive or snowball sampling is used. Participants must be knowledgeable about the topic and willing to talk; incentives and timing of recruitment matter.

6. Skill and role of the moderator — The moderator must be neutral, familiar with the topic, able to probe, control dominant speakers, encourage quiet ones, and avoid leading questions. A separate note-taker is desirable.

7. Venue and physical setting — A neutral, accessible, quiet place where participants sit in a circle; seating should not create hierarchy. Travel cost and convenience affect attendance.

8. Timing and duration — Sessions should suit participants' daily schedule (farm/office hours, festivals) and last about 60–90 minutes without fatigue.

9. Language and cultural sensitivity — Use the local language and culturally acceptable terms; respect taboos, gender norms, and local power structures.

10. Discussion guide (topic guide) — A semi-structured guide with open-ended questions and probes keeps the discussion focused while allowing flexibility.

11. Recording and documentation — Audio/video recording with prior consent, plus field notes on non-verbal behaviour; ethical approval and confidentiality must be ensured.

12. Ethical issues — Informed consent, voluntary participation, anonymity, and the right to withdraw; special care when discussing sensitive topics.

13. Data management and analysis — Transcription, translation, coding, and thematic analysis; group interaction itself is data, so notes on agreement/disagreement are important.

14. Resources and logistics — Budget, time, transport, refreshments, and manpower required for conducting and transcribing multiple sessions.

15. Researcher bias and reflexivity — Awareness of the researcher's own assumptions, and avoidance of dominating the discussion or steering responses.

Considering these factors ensures that FGD yields valid, reliable, and contextually rich qualitative data.

Significance of Formatting in a Research Report (5 marks)

Formatting refers to the systematic arrangement of the content of a research report — its title page, headings, sub-headings, tables, figures, citations, references, spacing, and pagination — according to an accepted style (APA, MLA, Chicago, etc.). Its significance for a good research report can be described as follows:

1. Ensures clarity and readability. Proper headings, sub-headings, margins, and line spacing guide the reader logically from one section to another. A well-formatted report is easy to read and understand, whereas a poorly formatted one confuses the reader even if the content is sound.

2. Maintains logical organisation and coherence. Formatting arranges the report into standard parts — preliminary pages (title page, certificate, acknowledgements, table of contents, list of tables/figures), main body (introduction, literature review, methodology, results, discussion, conclusion), and supplementary pages (references, appendices). This structure helps the reader locate information quickly and see the relationship among ideas.

3. Establishes academic credibility and professionalism. A neat, consistent format reflects the researcher's care, discipline, and seriousness. It creates a good impression on examiners, supervisors, and readers, and increases the acceptability of the report in academic and professional circles.

4. Facilitates proper citation and avoids plagiarism. Formatting prescribes standard rules for in-text citations and the reference list. This gives due credit to original authors, protects the researcher from charges of plagiarism, and allows other scholars to trace the sources used.

5. Aids effective presentation of data. Tables, graphs, charts, and figures are numbered, titled, and placed systematically under formatting rules. This makes quantitative findings attractive, comparable, and self-explanatory, strengthening the evidence base of the report.

6. Ensures uniformity and comparability. When reports follow a common style, different studies can be compared, evaluated, and reused by other researchers. It also makes the work acceptable to journals, universities, and funding agencies that insist on a prescribed format.

Conclusion: Formatting is not mere decoration; it is an essential requirement that gives a research report order, clarity, credibility, and communicative effectiveness. A report with excellent content but poor formatting fails to create the impact it deserves.

Importance of Literature Review in Scientific Research

A literature review is a systematic survey of existing knowledge on a topic. In scientific research it is not merely a formality but the foundation on which the entire study rests. Its importance can be highlighted as follows:

1. Identification of the research gap. By reviewing what has already been studied, the researcher discovers what remains unexplored. This gap becomes the justification and the problem statement of the new study.

2. Formulation of research objectives and hypotheses. Existing theories and findings guide the framing of precise objectives, research questions and testable hypotheses. In economics, for example, a review of consumption theories helps in specifying the variables and the expected relationships.

3. Theoretical and conceptual framework. The review supplies the concepts, definitions and theories (e.g., Keynesian, monetarist, new classical) that give the study its analytical base and help interpret results.

4. Selection of appropriate methodology. Earlier studies reveal which research designs, sampling techniques, models and econometric tools have worked, and which have failed. This saves the researcher from repeating known mistakes and justifies the chosen method.

5. Avoidance of duplication. It prevents the wasteful repetition of research already conducted and directs effort towards new contributions.

6. Comparison and validation of findings. The results of the present study can be compared with those of previous studies to confirm, contradict or extend existing knowledge, thereby adding to the cumulative body of science.

7. Identification of variables and data sources. The review points to the relevant independent and dependent variables, their measurement, and reliable sources of secondary data such as CBS reports, NRB publications and international databases.

8. Identification of inconsistencies and debates. Conflicting findings in the literature expose weaknesses in earlier work and open the door to refinement and new research questions.

9. Justification of the study's significance. A well-documented review shows that the research is timely, relevant and useful to policy makers, academics and practitioners.

10. Improvement of research quality and credibility. A thorough review demonstrates the researcher's command over the field, sharpens the arguments, and increases the acceptability of the study in the scientific community.

Conclusion. Thus, the literature review provides direction, theoretical grounding, methodological guidance and scholarly credibility to scientific research. Without it, research becomes blind, repetitive and weak in both design and contribution.

Contingency table: education level and employment status

A contingency table (cross-tabulation) shows the joint frequency of two categorical variables. Suppose 200 persons are surveyed, with $X$ = education level (Graduate, Postgraduate) and $Y$ = employment status (Employed, Unemployed).

Marginals. Row totals: 120 graduates and 80 postgraduates. Column totals: 130 employed and 70 unemployed.

Row percentages (conditional distribution).

Relationship. Postgraduates have a higher employment rate (75.0%) than graduates (58.3%). The unemployment rate of graduates (41.7%) is about 1.7 times that of postgraduates (25.0%). The data therefore suggest a positive association between higher education and employment.

Test of independence. Whether the association is statistically significant can be checked with the chi-square test:

$\chi^2 = \sum \frac{(O_{ij}-E_{ij})^2}{E_{ij}}, \qquad E_{ij} = \frac{(\text{row total})_i(\text{column total})_j}{N}$

For Graduate-Employed, $E_{11} = \dfrac{120 \times 130}{200} = 78$, against an observed 70. The other expected values are 42 (Graduate-Unemployed), 52 (Postgraduate-Employed) and 28 (Postgraduate-Unemployed). Then

$\chi^2 = \frac{(70-78)^2}{78}+\frac{(50-42)^2}{42}+\frac{(60-52)^2}{52}+\frac{(20-28)^2}{28} \approx 0.82+1.52+1.23+2.29 \approx 5.86$

With $(2-1)(2-1)=1$ degree of freedom, the 5% critical value is 3.84. Since $5.86 > 3.84$, we reject independence: education and employment status are associated in this hypothetical sample.