Human Capital_planning Workshop

Human Capital, Economic Growth and Total Factor Productivity in the Agricultural Sector Leonardo A. Lanzona, Jr. Professor Ateneo de Manila University

Objectives of the Paper To

provide an empirical evaluation of a simple (Solow) growth model in the case of Philippine agriculture. More importantly, to determine how the model can be improved by extending the model to include human capital, that is, by recognizing that labor in different sectors may possess different levels of education and different skills. To provide an alternative

Rationale  Three

main reasons for incorporating human capital in the study of productivity:

◦ Because human capital is correlated with savings and population, omission of human capital accumulation in productivity estimates will result in biased estimates of the impact of these variables on growth and total factor productivity ◦ Because of the productivity effects of human capital (noted in labor studies), for any given rate of human capital accumulation, higher saving or lower population growth leads to higher levels of productivity and income, and thus also human capital. Failure to consider the human capital will lead to misleading estimates of technical productivity. ◦ Because of the varied forms of human capital, differential effects of human capital should be included in the study of growth and productivity. Varied steady states can be achieved under varied human capital

A Preliminary Model Based

on Mankiw, Roemer and Weil

(1990) Suppose that output, Y, in an economy is produced by combining physical capital, K, with skilled labor, H, according to a constant returns, Cobb1 YK (AH) function Douglas production

where A represents labor-augmenting technology (or total factor productivity) that grows exogenously

Preliminary Model (cont.) Individuals

in this economy accumulate human capital by spending time learning new skills instead of working. Let u denote the fraction of an individual’s time spend learning skills and let L denote the total amount of (raw) labor used in the production in the economy. Unskilled labor learning skills over time generates skilled labor H in thefollowing manner: u

H e L

where is a positive constant. If u is zero, then , showing that all labor is unskilled.

Preliminary Model (cont.)  By

increasing , a unit of unskilled labor increases the effective units of skilled labor, H. In order to determine how much, we estimate

dlogH  du

 This

equation states that a small increase in increases H by the percentage . This formulation is intended to match the literature in labor economics that finds additional years of schooling increases the wage by something like seven percent.  In this model, u is assumed constant and given exogenously in the same way that the saving rate is also exogenous and

Preliminary Model (cont.) If

we let lower-case letters denote variables divided by the stock of unskilled labor, L, and rewrite the  1  k of (Ah ) production function inyterms output u per worker h  eas: where  % If we divide both sides y% of k the output equation byy%Ah, then . k% The variables and are sometimes referred to as "output per effective unit of labor“ and "capital per effective unit of labor.“

Preliminary Model (cont.) This

labeling is motivated by the fact that technological progress is laboraugmenting. Ah is then the "effective" amount of labor used in production. In the Solow Model, the capital  accumulation over time is seen as:

K  skY  dK

wheresk is the investment rate for physical capital and d is the constant  depreciation rate k  sy  (n  d)k Moreover, dividing by labor, where n is the growth rate of the labor force, the capital accumulation in per worker terms.

Preliminary Model (cont.) Incorporating

technology, we

 then have %

k  sky% (n  g  d)k%

% y steady state values of  % % 0 by k k variables and are found k sk setting which leads to  y n g  d .

The

 /1 Substituting* this sinto   the k

production

%  y

 n  g  d   function

Preliminary Model (cont.) Rewriting

this in terms of output per worker, we get: sk  y (t)    n  g  d   *

This



 /1

hA(t)

equation explains why countries are rich because (a) they have high investments in physical capital, (b) spend a large fraction of time accumulating skills, (c) have low population rates and (d) have high levels of

Computing for TFP Solving

we get



1

A fromy  k (Ah)  / 1

 y A    k

,

y h

With data on output per worker, capital per worker, and educational attainment for each farm, we can use this equation to estimate actual levels of A.



Problems (1) Estimates

of A computed this way are like the residuals from growth accounting: they incorporate any differences in production not factored in through the inputs. For example, we have not controlled for differences in the quality of educational systems across countries, so that these differences will be included in A. In this sense, it would be more correct to refer to these estimates as total factor productivity levels rather than technology levels

Problems (2) 

◦

◦

◦ ◦

Granger Causality Regressors could be correlated with the error terms and the results of the OLS might be biased by the endogeneity of the regressors Direction of causality among productivity and explanatory variables might also go from productivity to the explanatory variables. Usual econometric analysis might simply capture correlations, instead of causality. Solutions: 1. Use instrumental variables to control for the endogeneity 2. Carry out Granger causality tests in the error correction panel data, to detect possible reverse causality among the variables.

Policy Direction: Different Levels of Education An

attempt will be made to consider effects of different schooling levels: primary, secondary and tertiary. This can serve as a guide to policymakers in determining which types of training are important towards improving agricultural productivity.

Data Panel

data collected in the Bicol

Area; ◦ 1975, 1984 and 1994 surveys ◦ Production data on various agricultural products but more intensive data for rice ◦ Information available for various inputs ◦ Barangay data or policy variables are also available