1. Skip to navigation
  2. Skip to content
  3. Skip to sidebar

Economic Currents

Keep up to date with the latest UHERO news.

In Search of the Glass Ceiling: Deciphering Data on Gender and Wages

Posted January 27, 2014 | Categories: Blog

The gender pay gap gets a lot of attention, but what can we learn about it by looking at the wage data?

The Hawaii Equal Pay Dashboard compares female salaries to male salaries over time. It breaks up the workforce into a few dozen occupation groups, and uses American Community Survey (ACS) data from 2005-2012 to try to get a glimpse into our state's state of gender wage equality. The data however, presents a picture that's fuzzy and misleading to say the least. But it also helps to illustrate the complicated reality of wage inequality visualization.

For example, looking at the ACS median wage data we see that there are substantial differences across genders within the same occupation. However, the real big questions are: are these big differences really there? Do men in Legal Occupations really make 171% more than women in the same occupation? And do women really make 18% more than men in Installation, Maintenance and Repair Occupations?* The answers are not straightforward and are much more complex, a testament to the multi-faceted - and sometimes misleading - nature of data.

First of all, the ACS’s median wages for men and women are just estimates and the precision of these estimates varies across occupations. To produce an estimate, the ACS surveys a sample of housing units and people in the population, not the entire population. That’s why in addition to the median estimates, the ACS provides a 90% confidence interval, which is a statistical measure that reflects a degree of uncertainty around this estimate. The larger the confidence interval, the lower the precision of that estimate and the lower confidence we have that this estimate is close to the true value. For example, in the case of male salaries for Legal Occupations for the 2007-2009 aggregate, that confidence interval spans $116K - $180K, a huge range. It means that there is a 90% chance that the median wage falls within this range, but we’re not quite sure what exactly the median male in the legal profession earns.

Imagine telling someone that the geographically average person in the continental US lives in Chicago… but with a confidence interval that extends east to New York City and west to Los Angeles. What would we actually know about the way the population is distributed? It's difficult to tell, to say the least.

The size of the confidence interval is affected by the quality and the amount of data collected. One way the ACS is able to produce more accurate estimates for occupations with a small number of workers is by aggregating survey data over several years. Check out how the one-year aggregation data picture (data from just one individual year) compares to the three- or five-year aggregation data pictures (data created by averaging the results from clusters of three or five years, respectively). The confidence intervals narrow as more time periods are aggregated and more data are included in the estimate’s calculation. However, interesting characteristics of individual years are lost in the aggregation process.

Complicating things even further is the issue of important pieces of information not shown in this particular data view. An array of important factors such as education levels, experience, training, industry, job specific duties, and so on affect wages and can all affect this data-drawn picture of the state of equality. In a way, a comparison of the median wages within the same occupation without accounting for differences in qualifications, employment industry, and other factors that are responsible for differences in wages is like comparing papayas to taros.

The takeaway? While a glance at the tilt of the visualization gives us the impression that the data does seem to be skewing wages in the favor of men, the bottom line is that things might not be what they seem. To be able to make an accurate conclusion, the reader should be paying attention to what is displayed, and recognize the fact that visualizing equality is complicated. An enormous amount of additional information would have to be included in a visualization such as ours for it to really be able stand a chance at tackling the gender equality question in a meaningful way.

How do your real life experiences with the issue gender and wages in your occupation group compare? What other information would help you in understanding this issue better?


---Inna Cintina and Natalie Schack 

*This data is from the 3-year aggregate level for the years 2007-2008.

Are Recessions Bad for Your Health?

Posted January 21, 2014 | Categories: Blog

Our work indicates that a bad economy can kill you.

Specifically, we show that over the ten years from 1984 to 1993 that a one-percentage point rise in the unemployment rate increased the risk of dying within the subsequent year by 6% for working-aged men. This translates to roughly 24 more deaths per 100,000 people. A percentage point increase in the unemployment rate increases the number of unemployed by about 1.5 million, resulting in about 360 additional deaths of men between the ages of 30 and 60. Of note is that we find no such relationship for women or the elderly who tend to have weaker ties to the labor market.

Social scientists have long been concerned with the health consequences of business cycle fluctuations and, more generally, changes to one’s socioeconomic status. If public health tends to decline when the economy performs poorly then the costs of recessions are not limited to purely economic hardships such as unemployment and underemployment. Indeed, early work on the topic by Harvey Brenner, a public health researcher at the University of North Texas and Johns Hopkins University, suggested that mortality rates did tend to rise as the economy worsened. This work appears to indicate that poor job market prospects come with increased stressors that pose health hazards.

Subsequent work challenged these findings. Christopher Ruhm, a health economist at the University of Virginia, raised numerous methodological issues with the earlier studies and showed that once these were addressed mortality rates actually declined during recessions. This ushered in an era of work by health economists investigating if recessions were, in fact good for your health. The explanation for this counter-intuitive result was that people tended to lead healthier lifestyles during recessions. Recessions created more time for people to look after themselves and participate in health activities. Having less money also made people less likely to purchase harmful vices.

However, more recent work supports the earlier idea that recessions have a negative impact on public health. For example, research by Till von Wachter and Daniel Sullivan, economists at UCLA and the Chicago Fed, looked at micro-data from Pennsylvania and showed that involuntary job displacements during the early Eighties were associated with much higher mortality risks. Other work by researchers at UC Davis replicated Ruhm’s findings but showed that mortality rates also declined for the elderly and the young. Since those groups should have weaker labor force attachments, this research cast doubt on the healthy living mechanism.

So, what explains the difference between our finding and Ruhm’s? In principle, both of our studies should be identifying the same parameter and so should deliver the same or, at least, similar estimates. However, the one key difference is that our study uses data that are at the individual level, whereas other work uses data that are at the state level. This is noteworthy because higher levels of aggregation have been known to induce biases in primitive relationships that exist for individuals. Our paper argues that this use of individual data supports the case for the negative relationship between recessions and health.

While our work cannot tease out the mechanisms that generated our findings, we can offer some speculation. We believe the relationship between stress under poorer macroeconomic conditions and the risk of dying from cardiovascular-related causes may be a viable mechanism worthy of future investigation.

---Tim Halliday



UHERO 101.11: Hawaii Health Insurance Premiums, Oligopolies, and the Affordable Care Act

Posted December 20, 2013 | Categories: Blog

Earlier this year the Federal Department of Health and Human Services (FDHHS) issued a summary report on the health plan choices and premiumsacross the country that will be available in the Health Insurance Marketplace. This report focuses on assessment of the plans with the lowest premiums in each state, as those are expected to be popular among consumers. An interesting question is “Where do Hawaii plans/premiums stand relative to other states?” Unfortunately, there is not an easy answer to this question. The Hawaii Health Connector released its estimates too late to be included in the FDHHS report and the direct comparison of premiums across two reports is not straightforward. However, some useful observations can be drawn.

The FDHHS report notes that across 36 states included in the analysis, premiums tend to be lower in states where there is more competition and transparency. For years there have been talks about the level of competition in the Hawaii's health insurance market. According to the latest study done by the American Medical Association, Hawaii tops the list of states with the least competitive commercial health insurance market. This is a case where being at the top is not a good thing, at least not for prices.

“Oligopoly” is the way economists describe situations in which a particular market is characterized by a small number of firms that supply the entire market. A given industry with a large number of firms can be oligopoly as well, if a few of the firms are relatively large compared to the overall market (i.e. they produce most of the industry’s output)1. Two health insurance providers have dominated the Hawaii market for many years: the Hawaii Medical Service Association (HMSA) and the Kaiser Foundation Health Plan (Kaiser). Together, the plans claim nearly 90% of the insured, with HMSA representing more than 50%.

A small number of firms in Hawaii’s market does not necessarily mean relatively high premiums. The determination of premiums is an interesting and complex question. Premiums vary among states, reflecting variance in underwriting regulations, health care costs, demographic characteristics, and consumer preference. To illustrate, prior to the passage of the Affordable Care Act (ACA aka Obamacare), rules regulating the individual market for health coverage varied considerably across states. In some states (e.g., Maryland) companies can exclude individuals with pre-existing illness from coverage. This can help keep rates lower (but only for those who qualify). In other states (e.g., New York) the opposite rule applies: insurance companies were prohibited from excluding individuals with pre-existing illness. This prohibition has been associated with much higher rates in the market.

Based on a comparison of individual rates, the Hawaii Department of Commerce and Consumer Affairs concluded that the rates approved for the Hawaii Health Connector are among the lowest in costs in the nation. Without additional research, it's impossible to know why Hawaii's ACA plan costs appear to be lower than in some other states. It's worth noting that Hawaii’s market might, in some respects, be unique. The majority of the population here has health insurance. (The 2010-2012 estimates from the American Community Survey indicate that the proportion of people aged 18-64 without the health insurance coverage in Hawaii is 10%, whereas the corresponding nationwide statistics is twice as high at 21%. Among people 65 years and older it is 1.2%, about the same level as nationwide.) It's conceivable that lower overall premiums may be due to more effective risk pooling. Determining whether this is the case is perhaps the topic of future UHERO research.

--Inna Cintina


1  In such markets, competition between firms does exist as firms try to attract more customers, but it is realized via incentives rather than changes (decreases) in prices. Instead, prices are kept relatively constant, but firms engage in fierce advertising highlighting the differences across products to attract customers.

Lignocellulosic Ethanol - Are We There Yet?

Posted December 18, 2013 | Categories: Blog

Lignocellulosic ethanol has been touted as a cleaner, next generation alternative fuel. Turning abundant resources like grasses into a transportation fuel sounds like a good idea. But, is this a viable option in Hawaii? What's the state of technology now? What would it cost to produce it locally? Will it reduce Hawaii’s greenhouse gas (GHG) emissions?

The recently completed study titled "Market, welfare and land-use implications of Lignocellulosic bioethanol in Hawaiʻi " seeks to answer some of these questions. In this interdisciplinary study, bottom-up bioenginering and top-down computable general equilibrium (CGE) models are combined to estimate the likely cost of bioethanol production and its impact to Hawaii’s economy. The study assumes lignocellulosic bioethanol is produced via the state-of-the art Simultanerous Saccharification and Co-Firmentation (SSCoF) of Napiergrass, and is used to meet an alternative fuels standard (AFS) of 10% and 20% of the state’s gasoline demand, respectively. The four policy scenarios evaluated are: i) a federal blending tax credit, ii) a long-term purchase contract, iii) a state production subsidy financed by a lump-sum tax and iv) a state production subsidy financed by an ad valorem gasoline tax.

The modeling results indicate that Hawaii-produced bioethanol is relatively costly. Given the state of technology and economic conditions, all scenarios reduce local residents’ welfare (estimated to be around -0.14% and -0.32%). Unsurprisingly, Hawaii’s economy fairs best under the federal blending tax credit scenario: the policy support makes Hawaii’s ethanol competitive, and produces a positive impact to gross state product of $49 million. Otherwise, impacts on gross state product are estimated to be negative (up to -$63 million).

The study also finds that Hawaii-based bioethanol is unlikely to offer substantial GHG emissions savings in comparison to imported biofuel. The fuel tax scenario achieves the most GHG emissions reductions estimated at -2.3% (the 20% mandate) and -0.5% (the 10% mandate) of the total GHG emission in Hawaii. The federal credit case achieves the least reductions estimated at -1.4% (the 20% mandate) and -0.06% (the 10% mandate). The policy cost per tonne of emissions displaced ranges between $130 to $2,100/tonne of CO2e. The study also highlights that the value of pasture land could increase by as much as 150% in the 20% AFS scenario.

Overall, the study shows that technological and economic hurdles remain, but a number of emerging solutions could make this local option viable: a bioethanol plant may adopt an anaerobic digestion system, which may process local animal manure as an additional energy source of biogas. Further, a plant may produce an innovative fungal-based co-product to be used as a local animal feed substitute, such as those being evaluated at the UH MBBE department (Takara and Khanal, 2011). Also, the emerging concept of a biorefinery aims to produce a range of fuel products such as jetfuel and industrial chemical substitutes. Future studies are needed to evaluate such novel options in Hawaii.

--Junko Mochizuki and Makena Coffman 




Mochizuki J, Yanagida J. & Coffman M(2013) Market, welfare and land-use implications of Lignocellulosic bioethanol in Hawaiʻi. UHERO WorkingPaper No. 2013-10.

Takara. D., & Khanal, S. K. (2011) Green processing of tropical banagrass into biofuel and biobased products: an innovative biorefinery approach. Bioresources Technology 102(2); 1587-92. 




UHERO 101.10: The Confusing World of PV

This UHERO 101 intends to clarify some of the rate and policy aspects of PV in Hawai‘i, and explores the two opposite driving forces of PV adoption.

PV is an attractive investment in Hawai‘i where electricity rates are almost four times the national average. Rising electricity prices and falling system costs have largely driven the installation trend, with installations roughly doubling annually since 2007. Moreover, residential PV is quite cost-effective because it’s installation costs are up to 65% subsidized. In addition, there is ongoing support of PV in the form of Net Energy Metering (NEM). NEM gives households retail rate for their unused PV generation, rather than the wholesale rates paid for other sources of energy. As such, what many do not realize is that distributed PV can actually raise electricity rates rather than lower them. While PV customers benefit from providing their own energy and selling excess electricity back to the grid, non-PV customers are consequently likely to pay relatively more. Also, although PV certainly reduces the use of fossil fuels, it is not necessarily proportionately. Since PV is an intermittent source of energy, the utility also has to run spinning reserves to ensure reliable electricity at any given time.

To add to the confusing world of PV, Hawaiian Electric Industries recently modified its policy— both the primary metric used to determine circuit saturation and the process of connecting to the utility’s power grid. Prior to September 2013, distributed PV generation was limited to 15% of peak load on each circuit that, if exceeded, required the NEM applicant to pay for an interconnection requirements study (IRS). However this limit was not enforced for smaller systems under 10 kW (i.e. residential systems). However, now the metric of daytime minimum load (DML) is used to determine circuit saturation. The policy does not distinguish between small and large systems, and requires that written consent be obtained from the utility prior to installation.1  The details are summarized as follows:

     1.    Circuits below 75% of DML are not subject to an IRS or circuit upgrades. These projects should receive notice to proceed within 35 business days.
     2.    Circuits that fall between 75-99% of DML are not subject to an IRS but may require circuit upgrades. Depending on whether a supplemental review is required in addition to the initial technical review, these projects may receive a response anywhere from 35 to 85 business days.
     3.    Circuits beyond 100% of DML may require an IRS and circuit upgrades. Completing an IRS study may take up to an additional 165 calendar days on top of the initial and supplemental review.

As a result, on one hand, the policy change has slowed down solar installations, due to circuit upgrades. For projects above 75% of DML, customers have to first wait to hear whether a circuit upgrade is necessary and then, if deemed necessary, another several months for conducting the circuit upgrade. In addition to the long waiting period, potential customers face extra costs for circuit upgrades, which are allocated on a prorated basis and divided according to the size of the systems to be installed.

At the same time, the policy change provides further motivation for those customers who have been considering a PV system and whose homes are on circuits below 75% of DML, to join the race to install PV.2 

* If you are thinking of installing PV, as an initial circuit availability check, enter your address here: http://www.heco.com/portal/site/heco/lvmsearch 


Source: HECO


-- Sherilyn Wee and Makena Coffman


1In the past, submitting the NEM agreement was often the last paperwork step of the installation process. Applying for the City and County building permit was usually the first step, and now is applied for only after receiving approval to interconnect. Building permit approval takes about 20 business days.

2Previous state legislative discussions on reducing or phasing out the renewable energy investment tax credits have already commenced the “race” to install PV.

Page: 7 | 8 | 9 | 10 | 11 | 12 | 13 | 14 | 15