Economic Change in Ghana, Chris Udry Yale University
Goal: document trends in entrepreneurship and employment in preparation for Ghana panel survey and associated interventions Ghana LSMS 1987 through GLSS 2006; 5 rounds of comparable cross-sections 4,000 – 7,000 households
Trends in Ghana
Significant growth over 2 decades Marked decline in poverty – Dramatic increases in education – Urbanization – Health improvements, decline in birth rates What changes do we see in economic activity at the micro level? – New enterprises? – Enterprise growth? – Increased specialization?
Gradual Change in Primary Occupation
Gradual fall in Self-employment
Look at Investment patterns: Household Asset Portfolios, 1992
Declining share of business assets
For business owners, as well
Enterprise size over time MeansMedians year# of workers# of hh workers# of workers# of hh workers Enterprises are not growing By household, total enterprise employment MeansMedians year# of workers# of hh workers# of workers# of hh workers
Nor are they growing in terms of capital: Median kMeanMedian wealth
Evidence on Returns to Capital Farm profits in new technology Prices Enterprise profits
Prices of durable goods Prices of goods with varying life expectancies contain information on the discount rate
Can estimate this with data on prices and t (assuming measurement error in t) – used parts example Using this relationship, we estimate r=60%
NFEs earn high returns, particularly at low levels of K. - warning: sensitive to assumptions on w
GLSS 5 Return on Assets by Enterprise Capital Decile of Invested CapitalMedian Profit/K ^ n= 1707
Are business starts dependent on wealth?
Are commercial enterprises growing less, or not being started, due to capital constraints and risk? Experimental intervention associated with EGC surveys discussed earlier Driven by theory: role of risk and imperfect access to K on investment choice. Goal is to quantify effects of risk/risk aversion and capital market imperfections on investment choice
Not a policy evaluation; insurance is to be free Working with MIA to design an insurance product to address most salient dimension of risk of commercialization. Crop price or rainfall, likely rainfall for intensive maize. 2 x 2 design with grant experiment Sample of 500 farmers in high potential maize areas
Combine with data collection on expectations, and prospective uses of additional funds Simultaneous orthogonal evaluation of large- scale commercial training program for farmers
Annual Hours Worked
Labor Force Participation by Gender
Net effect, increase in work/capita
Determinants of Annual Hours Worked -- GLSS rounds 1,2,3, 5 variableestimatevariableestimate noschool18.05dist to road (30.83)(1.979) primary66.95months road passible0.627 (35.23)(3.256) jss105.3bank distance (29.81)(0.578) sec88.71post distance1.003 (39.49)(0.668) male = daily market in community (17.26)(23.89) age48.05any market in community11.01 (2.027)(18.65) age squared-0.527restaurant in community67.08 (0.0232)(20.21) urban346.2y (19.26)(21.73) electricty6.669y (27.88)(22.30) pipe or borehole water-72.91y (24.15)(19.76) Male*electricity-27.32Constant550.9 (35.78)(55.59) Male*water (34.97) Observations25880 R-squared0.140
No increase in specialization: