Tuesday, 30 June 2009

Technology epidemics

When a new technology is invented, the number of people who are using it is often small for a while, then increases quickly, then slows down. Some earlier posts showed the emergence pattern in different countries of ICT technologies (for example, here).

There are competing explanations how the characteristic S-curve arises. One is that people start using the technology as they are exposed to it by other users. When there are few people using the technology, not too many people are exposed to it and it spreads slowly. When more people are using it, exposure is greater and spread is faster until almost everyone is using it and so spread slows again. Other explanations are based on the speed of the technology's acceptance or to its skill requirements and the distribution of people's skills.

The explanations give the same predicted curves, but the implications for technology policy are different. If technology spread is limited by exposure, then a policymaker may best promote it by publicity or other measures to increase knowledge of it. If on the other hand, spread is limited by skills, then training is the best way of promoting it.

For readers interested in the mathematical models, Wikipedia has a page on them here. The derivations corresponding to each of the above explanations would be a little different, but the outcomes are similar. The models are also used in modelling disease spread, so are very useful for comparatively little work.

Free good-quality maths software

Maxima (available at no charge here) does maths analysis, and lots of it: integration, differential equations, simplifications, graphs, roots, and more. Thanks to the public spirited providers. There is science and other software on the host site, SourceForge, but I have only looked at a fraction of it to date.

Contender for charity of the year - Toilet Twinning

A strong contender for charity of the year is Toilet Twinning (here). You pay £60 and your toilet is twinned with one built with the money in Burundi. You get a picture of the unique latrine. That's it.

The project is run by a major UK charity, so presumably its value-for-money is comparable to other aid projects. But even if the money was being burnt; even if it was being used for palaces; look at the logo...

I have no affiliation with Toilet Twinning, and they had no connection with this post.

Friday, 26 June 2009

Pressing the economic spring in Eastern DR Congo

I wrote in a previous post that Western DR Congo could and should develop economically as rapidly as possible despite the fighting in Eastern Congo, even if no adequate solution can be found for the East. The advice leaves open the question of what to do there.

I have been thinking about the description in my recent post on output (here) of how some economies were like coiled springs when the government enforced conditions that varied from growth maximisation. The economies had high levels of technological knowledge or education, but restrictions on accumulation. When the economies transitioned to an open capitalism, they experienced rapid growth.

There is a parallel between these countries during their pre-maximisation stages and Eastern Congo, in that private accumulation of fixed physical assets is very difficult there because of the conflict. So preparatory human capital and technological skill accumulation might be a way of preparing the economy to bound forward after the end of conflict.

My idea is not advanced at the moment, and I am trying things out. I think that it might turn on flexible education, creating the conditions to promote such education, and exposure to international technology.

Is female parliamentary representation or professional representation more associated with greater gender equality?

My previous post (here) looked at the UN's gender equality measure (GEM) and its components. Across countries, there was not a strong relationship between parliamentary representation of women and their representation in professional and technical jobs. I decided to look at the relation in more detail.

The first graphics show the ratio of women's share in parliamentary seats to their percentage representation in professional and technical jobs. For ease of representation, I have split the data into two graphs, showing high and low ratios. At the top of the list comes highly gender equal countries in Northern Europe, as measured by the GEM, highly unequal countries in South Asia, and a scattering of other countries including the two East African countries of Tanzania and Ethiopia. At the base of the list is low gender equality former socialist countries and Middle Eastern countries.

The third graph shows the ratio plotted against the GEM measure. As the ratio of parliamentary representation to professional representation increases, gender equality increases but more slowly than the ratio.

The relationship between the ratio and GEM appears to be non-linear, so I took logs before regressing GEM on the ratio. Including the three South Asian countries of Pakistan, Nepal, and Bangladesh weakened the relation considerably and were dropped. The outcome was a good-fitting relation

ln(GEM) = -0.12 [0.06] + 0.40 [0.06] * ln(parl/prof) + robust error
(R^2=0.61; n=78; s.e.s in brackets)

or GEM = 0.89*(parl/prof)^0.4, with some error.

The regression and graphs raises some questions:

1. if there is a causal link for parliamentary representation increasing gender equality more than professional representation does, or

2. if it is the other way round, so gender equality brings increased parliamentary representation above increased professional representation,

3. why East Africa (and into the Great Lakes regions too) have high levels of female parliamentary representation,

4. why in South Asia parliamentary representation is associated with less gender equality than elsewehere, and

5. why former communist countries have far more professional equality than political equality.

[A technical clarification on spurious regressions:
I thought a little about the estimation ln(GEM) = -0.12 [0.06] + 0.40 [0.06] * ln(parl/prof) + robust error. There is a technical issue that many readers may wish to ignore, but I should really clarify to avoid misleading impressions.

GEM has a form (approximately) like parl+prof+another term. ln(GEM) may be roughly approximated as parl+prof+other term. ln(parl/prof) may be roughly approximated as parl-prof. Thus we have a regression that looks a bit like parl+prof = a + b*(parl-prof). Now if parl and prof are independent, we would have a regression estimate for b (assuming parl and prof are zero mean to simplify the algebra) of sum((parl-prof)*(parl+prof))/sum((parl+prof)^2). Taking expectations and using limiting theorems we have b=sum(E(parl^2)-E(prof^2))/(positive number). If E(parl^2) does not equal E(prof^2) then we obtain a positive coefficient for b. It has arisen solely by virtue of the algebraic manipulations used; any two variables parl and prof would serve equally well.

If there is a further relationship between parl+prof and parl-prof not produced solely by manipulations, then by the reverse argument we would have parl and prof not independent (perhaps this statement could be made more precise). Now the observation made in the earlier post was that there is not a strong relationship between the two; in fact correlation is very low. But looking at their graph, there seems to be a relationship over much of the variables' domains. So the variables do not seem to be independent, although having a low correlation. Thus, the relationship between ln(GEM) and ln(parl/prof) does not seem to be spurious.]

freerice.com expands to foreign languages, maths, science, and art

When last featured on this site, freerice.com was giving charitable donations of food if visitors to its site performed well at a quiz. Back then, one had to say what English words mean. Now the site has been expanded to quizzes about other languages, maths, chemistry, and art. I like, j'aime, yo gusto.

Monday, 22 June 2009

Changing food production techniques

For the food production techniques in a country to change, its food producers must know about the possibility of changing, have access to the alternative techniques, and decide that the benefits of changing are greater than the costs. The costs of changing are often discussed in the literature on technology transfer. Features that are relevant to costs include:

- whether the new techniques are suitable for the company's available productive resources (for example access to water),
- how much finance and capital is required (techniques requiring very large-scale production may be unsuitable for instance),
- how much knowledge is required and its learning costs,
- how geographically distant the company is from expertise about the technique,
- how much support and advice it can get,
- whether the techniques are associated with the use of commercialised or free products (for example in using genetically modified seeds),
- and whether the techniques are socially suited to the area (for example meat rearing in a certain way may have social meaning).

For a private company, the benefits can be measured as income, which depends on end product market demand. When technology transfer research examines the product market, it is often by assuming a simple form of substitution between the new technology's product and the old technology's product. For particular forms of food, more sophisticated modelling of the end market is possible. The demand for the end product is influenced by factors including:

- price,
- the exact type of the product (for example the type of coffee),
- preparation given to the product (for example by cutting into chunks before selling).
- nutritional content,
- safety (some foods being associated with more health risks than others),
- public information about the product's benefits or risks,
- appearance,
- perceptions of naturalness and wholesomeness,
- advertising and branding,
- place of consumption (for example in a restaurant or at home),
- social context (some foods may be considered suitable in certain circumstances, for example holidays),
- familiarity,
- habit,
- and population demographics.

The chances of a technique being successful in a country depends on the cost and benefit factors being favourable for profit. Other things being equal, shifting a single factor in favour of the technique will make it more likely that it is successful in the country. So a technique where the knowledge requirements are not excessive in the country is more likely to be successful than one where great knowledge advances are required. A technique which does not create health concerns in the public is more likely to be successful than one that does. Countries have different circumstances, and no single technique can be successful in every country. It matters to choose techniques well.

Localisation of spillovers of gender equality, by component

A recent post here looked at whether gender equality spills over international borders, where equality is measured by the UN's gender empowerment measure. The finding was that it does a little, but the effects fall off quickly with distance. National income and other factors are much more important than international spillovers.

Today, I split up the measure into its components: the percent of seats in parliament held by women, the percent of professional and technical workers who are female, and the percent of professional and technical workers who are female. The GEM measure corrects for the proportion of the population who are female, whereas the plain components I use do not. Then the same estimations were run as in the previous post. The table shows the results.

The spillovers from international equality are weak for the components. The strongest spillover is among levels of equality in professional and technical work, perhaps because the factors embodying business institutions are more substitutable across countries than the factors embodying parliamentary institutions - a businessperson may find it easier to be a businessperson in a neighbouring country than a parliamentary representative would to switch parliaments, for example.

The explanatory power of the component regressions drops sharply compared with the full GEM model. Only parliamentary representation is explained to any extent by the model. Part of the drop is probably due to the correction for female proportions in the full GEM measures. A bigger part is plausibly caused by the weak correlation of the three components across countries. A country with a high professional equality has almost no tendency to increased parliamentary or legislative equality. Equality in one country in one component of the GEM may be related to equality in a different component in a different country.

Surfing Liberia

Here is a pretty photoreport on surfing in Liberia. The Liberian coast may be able to position itself as a surfing destination for tourists or locals, if peace allows. The World Bank has published work on tourism development in Africa (for example here).

Friday, 19 June 2009

How much is my country's capital worth?

Capital is the physical equipment and material used in producing new goods. It is made by investment. Much research has found a link between a country's capital and its output, and so there is a link with investment too. Investment is something that can be changed by a country, and so other things being the same, a country can alter its output too.

If output grows quickly when capital increases, then investment is even more important. To find out how strong the link is, we have to have good measures of output and capital. Output is relatively easy to work out since much of it is sold every year so it can be found by surveys, amongst other means. For capital, it is trickier, since it is not completely sold every year.

One method of finding out the capital in a country is to ask everyone and every company what equipment they have, and then work out its value by looking at the market prices for the equipment, if it is still sold. The method will be time-consuming and expensive, if it is possible at all.

An alternative method is to say that investment makes capital, so to estimate a country's capital we can add up its past investment in some way. Many writers assume that investment goes straight into capital, so 10 francs of investment become 10 francs of capital. They also assume that past investment becomes less valuable every year. 10 francs of capital made this year are worth more than 10 francs made a year ago, which is worth more than 10 francs made two years ago.

Old capital may be less valuable because it breaks. The same equipment in two different countries may break at similar rates. It may also become less valuable because it is used to make a product that fewer people want to buy as time goes on. People's preferences may change faster in one country than another if the country has faster improvements in the products it has available. These countries are likely to have high investment rates, too, since high investment is likely to encourage technological change. So high investment is likely to be associated with faster reduction in capital value.

Sometimes, someone might be interested in how fast a country's capital deteriorates physically. At other times, they might be interested in how valuable it is for making new goods. Deterioration rates and whether they are the same across countries depend on how people want use the calculated capital.

There is a research paper discussing these issues here.

People, bananas, dogs, and QSAR

A handful of scientific articles and papers caught my attention in recent weeks.

Population density triggers cultural explosion (here)
The work fits neatly with recent economic studies on localization of technological spillovers and output benefits of employment density (for example, here).

An account of the diseases striking the world's favourite fruit (here)
Bananas are a staple food for many people in Africa, and subject to potentially terminal diseases.

Comparison of dog genes (here)
I asked a while back why dog breeds look dissimilar and cat breeds similar. The paper presents a partial answer, showing gene differences in dog breeds, and raises a dozen further questions.

The quantitative structure-activity relationship (here)
QSAR examines the relationship between a molecule's structure and its chemical properties. I found it interesting in relation to my work on technology, as it helps to show how a company's research can transform into potentially commercialisable property. I wonder if there are similar innovation tools in the other sciences.

Monday, 15 June 2009

Three types of graph commonly used in business and government reports

Some people prefer words to numbers in describing how economics works. Sometimes words are clumsy for description, and graphs are helpful to represent their economic information. A few types of graph occur frequently in academic, business, and government reports, and it is worth being familiar with them.

Scatter graphs (also known as X-Y graphs)
Suppose we have received some data that says that when an economy produced $1000 million of goods, imports were $100 million. When the economy produced $2000 million, imports were $400 million. When the economy produced $3000 million, imports were $1000 million. Looking at the data, we could say that as the economy became larger, imports increased.

We could alternatively draw a type of graph using the data, known as a scatter graph. We draw evenly spaced points on the line at the bottom of the graph, called the x-axis, including the points from the data describing production. We draw evenly spaced points on the line at the left side of the graph, called the y-axis, including the points from the data describing exports. We have joined up the points with a line.

The graph shows that as production increases, so do exports, which is what we said earlier. We can also see that as production grows, exports seem to be getting larger more quickly. We can also guess at what exports would be if the economy produced other amounts of goods, by seeing where the line goes. For example, when production is $1500 million, we can estimate that exports would be about $230 million.

Bar charts
Suppose that we have been told a little more information, that the country producing $1000 million of goods is called Angoland, the country producing $2000 million is called Beninia, and the country producing $3000 million is called Congoroon. We can draw a graph with the countries along the x-axis and production on the y-axis. For each country, we draw a column up to their production. The graph looks like this:

We can see quickly from the graph that Congoroon has the largest production, while Angoland has the smallest. Bar charts can be particularly useful if we have much disorganised data. We can see quickly which countries are the largest, smallest, or near the middle of the data.

Pie charts
Pie charts are another way of presenting the information in the bar chart. Angoland produces a sixth (1000/6000) of the total production of the three countries, Beninia produces a third (2000/6000), and Congoroon half (3000/6000). A circle is drawn, with a sixth of the area marked by Angoland, a third by Beninia, and half by Congoroon, like this:

Pie charts allow us to see quickly how important a country is for production and compare countries, which might not be easy from the numbers.

Friday, 12 June 2009

Income elasticity of ammonia imports in SSA and developing Asia

Here is a test of the importance of industry in developing Africa and Asia. Ammonia is a chemical with wide applications throughout industry, so if a country is substantially altering its industrial production the amount of ammonia it imports is likely to change too. I ran regressions to estimate the income elasticity of ammonia imports (a measure of the change in ammonia imported as income increases) for five industrialising Asian countries and five African countries using annual data from here. The results are in the table. Changes in income generally explain a far higher amount of ammonia import changes in the Asian countries than in the African countries, as measured by the R2 statistic.

The lowest income Asian countries (Cambodia and Vietnam) saw their ammonia imports fall steeply as they became richer. By comparison, the only African country with a strong link between income and ammonia imports (Kenya) has a positive link. A tentative explanation is that the two Asian countries switch to manufacturing their own ammonia imports as they industrialise, while Kenya continues to use foreign imports. Another explanation is that there are different shifts in and out of processes using ammonia across the countries. It would be interesting to know which explanation applies. There might be a market for African chemical manufacturers, given that ammonia manufacture is a well-known procedure (described here).

How should a government invest to maximise national output?

Suppose a government has a budget and wants to invest in the economy to maximise economic output. It can choose between educating its population, buying more of its current productive goods such as roads or factories for state owned companies, or researching for new technologies and adopting foreign technologies. Which should it do?

Here is a quick answer that illustrates some of the major principles involved. Let us say that the economy's output is given by a function

Output = Y = A^a * K^b * H^c.

This says that output depends on the technology used in the economy (A), the invested money in physical inputs (K), and the education in the economy (H). The exponential terms a, b, and c describe how output may rise at a different speed to the inputs.

Suppose that the costs of A, K, and H are constant relative to each other, which may be the case if the economy is small relative to the input sources (such as the rest of the world) or if the factors used to make A, K, and H (such as finance and labour) are interchangeable. Then the country has a budget


for some total amount of money f, where d and e are the prices of K and L relative to A. There are several ways for the government to work out how to maximise output when its budget is constrained. One way is to substitute the budget into the output function and then use differential calculus to find a maximum. Another way is first to take logarithms of the output function and then substitute the budget, which is less algebraically complicated, and equivalent since the logarithm function moves up and down at the same places as the original function so will have the same maximum. Another way is to use linear programming, subtracting an undefined multiple of the budget from the output function and then using the Lagrange multiplier approach (described here. This approach is very common in academic work since the algebra tends to be by far the least complicated for big problems).

The solutions are

K/A = b/(a*d) for all f and L/A = c/(a*e) for all f.

So the ratios of the inputs at the maximum output are constant. Investment should occur to keep these ratios constant. The constants a, b, c, d, and e can be estimated from past data, although the estimation might be difficult since: 1) the economy may have changed since the data was generated, 2) the economy may not have a maximal allocation in the past, 3) the data should be for government expenditure, not for economy-wide expenditure and the data may not be available, 4) the assumptions of parameter stability are only approximate, 5) the function for output is an approximation, and 6) there are other reasons too.

The many cautions on estimation just presented show why the answer is rough, but it does have a place when there is an extreme imbalance between the amounts of each input (far more capital in the economy than technology, for example), since then the approximations are less likely to make the broad recommendations wrong. Large imbalances are not uncommon internationally; communist countries often have far higher levels of education than technology for example, so increasing expenditures on adopting new technologies is more likely to promote growth than further educational expenditures. Another example might be in closed economies that tried to import-substitute for advanced Western goods in the past; they built up large technological bases (even if the technology was difficult to observe because of limited capital stocks) and so capital investment is more likely to promote rapid growth than further domestic technological innovation. Such economies are like coiled springs, where political decisions caused deviation from growth maximisation in the past, but were able to correct rapidly to their long-run position on adoption of a growth-maximising approach to the economy.

Economic value added

Economists like to specify production functions that say output depends on various inputs such as capital and labour. They may write Y = F(K,L) or Y = Y(K,L). They also specify a budget, stating how much the inputs cost and how much money there is to buy them, such as a*K + b*L = total funds for purchase.

The output Y is different from the cost of inputs. The difference between the two is the value added in the economy or company producing the output. Value added arises from the organisation of the inputs to produce the output (some economists refer to the organisation as technology, although others keep the term technology for a particular part of the organisation). The value added is taken by the organisers (who may also provide the capital and labour). If Y is measured by K multiplied by L, say, then although the production function mentions only K and L there is still organisation used as the value added is greater than zero. The only time that there is no organisation is when the output equals the cost of the inputs.

Tuesday, 9 June 2009

Applications for MA Economic and Governmental Reform at the University of Westminster

Here's a reminder about applying and getting funded for the Master's course in Economic and Governmental Reform at the University of Westminster here in London, starting in October. I teach the economics modules on the course. African applicants are most welcome and have good performance records.

Our students have come from government, private sector, and NGO backgrounds, and after the course have moved on to senior positions in Africa, Europe, and beyond. Living in London itself offers many attractions and opportunities, of course.

Information on the course and obtaining funding is on its website (here). The course, like most in the UK, is expensive (GBP10,000), so students usually have applied for scholarships first. Course requirements are listed on its website, although there is some flexibility. Unavoidable ones are:

1. Reasonable English (or things won't make sense)
2. A first degree with some relevance to the topic, or a degree and relevant work experience
3. Willingness to work hard (or things will not be enjoyable)

Good luck with applications.

Monday, 8 June 2009

Is gender equality subject to localised international spillover?

I suggested in a post last week that institutions could be considered as technologies based on people. They may inherit some of the properties of technologies, and so today I considered whether institutions in one country tend to affect the institutions in other countries and whether the effect is altered by the distance between the countries. I examined a single institution, a country's gender equality, measured by the GEM indicator from the United Nations described here (in a 5.5 MB pdf document on pages one and six).

The equation I looked at was

where the suffixes denote the value of the quantity in country i or c. denotes the distance between country i and country c. The population multiplier allows for greater effects from larger countries.

The GEM indicator is available here. The distance data is from here, and is in the form of population weighted distances between two countries including internal distances (in tens of thousands of kilometres). Population (in tens of millions) and gdp per capita (in thousands of current PPP US dollars) is from UN sources here.

A cross section of countries was considered for the year 2007: Argentina, Australia, Austria, Bangladesh, Belgium, Belize, Botswana, Brazil, Bulgaria, Cambodia, Canada, Chile, China, Colombia, Costa Rica, Croatia, Czech Republic, Denmark, Dominican Republic, Ecuador, Egypt, El Salvador, Estonia, Ethiopia, Finland, France, Georgia, Germany, Greece, Honduras, Hungary, Iceland, Iran (Islamic Republic of), Ireland, Italy, Japan, Kazakhstan, Kyrgyzstan, Latvia, Lithuania, Malaysia, Mauritius, Mexico, Mongolia, Morocco, Namibia, Nepal, Netherlands, New Zealand, Norway, Pakistan, Panama, Paraguay, Peru, Philippines, Poland, Portugal, Russian Federation, Saint Lucia, Singapore, Slovakia, Slovenia, Spain, Sri Lanka, Sweden, Switzerland, Thailand, The former Yugoslav Republic of Macedonia, Trinidad and Tobago, Turkey, Ukraine, United Republic of Tanzania, Uruguay, Viet Nam, and Yemen. Estimation was by non-linear least squares with bootstrapped standard errors.

The results are shown in the table. P values are in brackets.

The first specification finds that a country's gender equality is positively linked with other countries' gender equality, with the link diminishing with distance. The geographic localisation is quite strong; the chi parameter implies that the effect of a country's GEM on another country's GEM halves over 320 kilometres. The sharpness of the decline suggests that interpersonal movements may be important in the link.

When GDP per capita is introduced as a control in the second specification, the intercountry link becomes highly insignificant. In specification three with only GDP per capita as an explanatory variable, the R2 is almost the same as in specification two, indicating that the effect of intercountry links are substantially captured by GDP per capita. The explanatory power of the regressions involving GDP per capita are also far higher than for the regression involving only the intercountry links. The evidence supports gender empowerment as being much more directly linked to internal GDP per capita than external spillovers. I have not attempted to find the direction of causality in any links between intercountry spillovers, GDP per capita, and gender empowerment.

As an application, a very gender equal country is unlikely to have a major influence on regional equality except insofar as it boosts income in its neighbours. It would be interesting to see whether other institutions such as democracy are determined far more by internal conditions than external spillovers.

Thursday, 4 June 2009

The locus of problem solving, search engines, and African marketplaces

From the late 1980s onwards, several high-profile research papers appeared looking at how businesses can best serve buyers when the businesses and clients do not know exactly what the other one knows or wants. When this happens, businesses may offer buyers a product or service that is not what the buyer wants at all. To get around the difficulty, businesses and buyers may work together on different parts of the product design, for example by businesses dealing with the parts they know best such as engineering processes and buyers dealing with the parts relating to specific design. The paper here discusses the issues and gives examples.

The research could have been used to anticipate the success of internet search engines. People know what they want to find out, and the search engine knows how to find it. The search engine does not provide a list of things the person might want to find without first asking them. The most successful search engine, Google, has a plain main page, for example. Search engines differ in the degree to which they equip the user for the second stage of user input, where the user clicks on a link to visit a website. I have mentioned the World Bank's basic search engine before on this blog, for providing not-very-helpful search results. The newly launched Bing search engine looks like it is making extra effort to provide the user with information so that the user's second stage of input is made easier.

The analysis can be applied to African marketplaces too. For example, a market chaotically arranged with no indication of where to find goods provides little scope for buyers to use their known purchase preferences. A market may provide a map or billboard with the positions of stalls indicated; buyers can use their knowledge to a greater extent but there might be limitations on them if the map is incorrect, or they cannot read well, or the map location is in an unsafe area. A market designed to give buyers the best opportunity for using their information might have pictures of the product types with arrows pointing clearly and accurately in the direction of the stalls, with the pictures displayed prominently throughout the market.

Distance, technologies, and institutions

Recent work on how technologies such as computers spread across countries has found that a country's productivity benefit derived from foreign technologies tends to fall off sharply if the source of the technology is far away. The effect is possibly due to lower trade or movement of people between distant countries compared with closer countries. Here is one paper finding such results. They depend on how much a country is able to absorb technology, for example whether it has a suitably educated workforce. The findings and their interpretations are still in development and there is uncertainty about them at the moment.

I suggested in a recent post that institutions could be considered as technologies based on people. The analogy raises the question of whether institutions have a similar spread over distance. Some of the diffusion mechanisms are the same and hence may display the same decline with distance. The distance effect on institutional diffusion is suggested by recent tendencies in democratisation after communism (Eastern Europe being most liberalised with decreasing liberalisation towards Asia), in conflict (spilling out of Rwanda, Uganda, and Sudan and into neighbouring countries), and in religion (with countries furthest from the Arabian Peninsula having less politicised religion).